From caa596ab3f0f8768ee326d6e3d5d39782194676c Mon Sep 17 00:00:00 2001
From: Kakaru <97896816+KakaruHayate@users.noreply.github.com>
Date: Wed, 29 Jul 2026 20:27:35 +0800
Subject: [PATCH 01/30] ggml-cuda : disable MMQ on devices with less than 48
 KiB shared memory (#26141)

ggml_cuda_should_use_mmq() selects MMQ purely from the quantization
type. The current MMQ configurations are designed and maintained against
a minimum of 48 KiB per-block shared memory, the limit provided by
NVIDIA Pascal GPUs and later. On devices that report less, no supported
MMQ tile fits and mul_mat_q_switch_J() aborts when every tile size
exceeds the device's per-block shared memory budget.

Disable MMQ when smpbo < 48 KiB so the caller falls back to the BLAS
path instead of hitting GGML_ABORT. Some current MUSA QY1 devices
report only 28 KiB and are covered by this guard.

Reproduced on a Moore Threads MTT S70 (arch mp_21, 28 KiB shared memory
per block) with an RWKV-7 0.1B Q8_0 model:

  $ llama-bench -m rwkv7-g1d-0.1b-Q8_0.gguf -p 128 -n 0
  J_best=0
  ggml/src/ggml-cuda/template-instances/../mmq.cuh:1521: fatal error
  (core dumped)

Only prefill (batch > 1) is affected; token generation is fine. After
the fix the same device falls back to the BLAS path:

  Q8_0    pp128 1470.7 t/s, tg8 55.3 t/s   (was: abort)
  FP16    unchanged
  Q4_K_M  unchanged

This matches a -DGGML_CUDA_FORCE_CUBLAS=ON build (pp128 1464.2 t/s),
which confirms the fallback path is the one being taken.

This is not MUSA-specific: any device with less than 48 KiB per-block
shared memory is affected.

Co-authored-by: KakaruHayate <KakaruHayate@users.noreply.github.com>
---
 ggml/src/ggml-cuda/mmq.cu | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/ggml/src/ggml-cuda/mmq.cu b/ggml/src/ggml-cuda/mmq.cu
index 8a0f4d3b5..8cf1a3eb9 100644
--- a/ggml/src/ggml-cuda/mmq.cu
+++ b/ggml/src/ggml-cuda/mmq.cu
@@ -296,6 +296,15 @@ bool ggml_cuda_should_use_mmq(enum ggml_type type, int cc, int64_t ne11, int64_t
         return false;
     }
 
+    // MMQ tiles require at least 48 KiB per-block shared memory; fall back to BLAS otherwise.
+    {
+        const int    id    = ggml_cuda_get_device();
+        const size_t smpbo = ggml_cuda_info().devices[id].smpbo;
+        if (smpbo < 48 * 1024) {
+            return false;
+        }
+    }
+
     if (turing_mma_available(cc)) {
         return true;
     }

From afeebe103bd99cda8f5dfaefcabadf890db7fda7 Mon Sep 17 00:00:00 2001
From: Xuan-Son Nguyen <son@huggingface.co>
Date: Wed, 29 Jul 2026 18:02:30 +0200
Subject: [PATCH 02/30] llama: move suppress_tokens handling to common/sampling
 (#26276)

* llama: move suppress_tokens handling to common/sampling

* address security issues

* rm has_logit_bias
---
 common/common.h       |  4 ----
 common/sampling.cpp   | 15 +++++++++++++--
 include/llama.h       |  3 +++
 src/llama-vocab.cpp   | 17 ++++++++++++++++-
 src/models/gemma4.cpp | 37 -------------------------------------
 5 files changed, 32 insertions(+), 44 deletions(-)

diff --git a/common/common.h b/common/common.h
index e7c55ae92..14889193d 100644
--- a/common/common.h
+++ b/common/common.h
@@ -294,10 +294,6 @@ struct common_params_sampling {
 
     bool backend_sampling = false;
 
-    bool has_logit_bias() const {
-        return !logit_bias.empty();
-    }
-
     // print the parameters into a string
     std::string print() const;
 };
diff --git a/common/sampling.cpp b/common/sampling.cpp
index 7b241e34f..256ac161e 100644
--- a/common/sampling.cpp
+++ b/common/sampling.cpp
@@ -310,8 +310,19 @@ struct common_sampler * common_sampler_init(const struct llama_model * model, st
         }
     }
 
-    if (params.has_logit_bias()) {
-        samplers.push_back(llama_sampler_init_logit_bias(llama_vocab_n_tokens(vocab), params.logit_bias.size(), params.logit_bias.data()));
+    // logit bias: user biases + model suppress tokens (-INFINITY)
+    {
+        std::vector<llama_logit_bias> merged = params.logit_bias;
+
+        int32_t n_suppress = 0;
+        const llama_token * suppress = llama_vocab_get_suppress_tokens(vocab, &n_suppress);
+        for (int32_t i = 0; i < n_suppress; ++i) {
+            merged.push_back({ suppress[i], -INFINITY });
+        }
+
+        if (!merged.empty()) {
+            samplers.push_back(llama_sampler_init_logit_bias(llama_vocab_n_tokens(vocab), merged.size(), merged.data()));
+        }
     }
 
     if (params.mirostat == 0) {
diff --git a/include/llama.h b/include/llama.h
index 3c6d22be8..a3ceecf5e 100644
--- a/include/llama.h
+++ b/include/llama.h
@@ -1102,6 +1102,9 @@ extern "C" {
     LLAMA_API bool llama_vocab_get_add_eos(const struct llama_vocab * vocab);
     LLAMA_API bool llama_vocab_get_add_sep(const struct llama_vocab * vocab);
 
+    // model-specific suppress tokens (gguf key: tokenizer.ggml.suppress_tokens)
+    LLAMA_API const llama_token * llama_vocab_get_suppress_tokens(const struct llama_vocab * vocab, int32_t * n_suppress_tokens);
+
     LLAMA_API llama_token llama_vocab_fim_pre(const struct llama_vocab * vocab);
     LLAMA_API llama_token llama_vocab_fim_suf(const struct llama_vocab * vocab);
     LLAMA_API llama_token llama_vocab_fim_mid(const struct llama_vocab * vocab);
diff --git a/src/llama-vocab.cpp b/src/llama-vocab.cpp
index 9164a4dd8..443cd4640 100644
--- a/src/llama-vocab.cpp
+++ b/src/llama-vocab.cpp
@@ -2578,7 +2578,14 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
             if (suppress_idx != -1) {
                 const int n = gguf_get_arr_n(ctx, suppress_idx);
                 const int32_t * data = (const int32_t *) gguf_get_arr_data(ctx, suppress_idx);
-                suppress_tokens.assign(data, data + n);
+                // drop out-of-range ids
+                suppress_tokens.reserve(n);
+                for (int i = 0; i < n; ++i) {
+                    const int32_t id = data[i];
+                    if (id >= 0 && id < (int) id_to_token.size()) {
+                        suppress_tokens.push_back(id);
+                    }
+                }
             }
         }
 
@@ -4205,6 +4212,14 @@ bool llama_vocab_get_add_sep(const struct llama_vocab * vocab) {
     return vocab->get_add_sep();
 }
 
+const llama_token * llama_vocab_get_suppress_tokens(const struct llama_vocab * vocab, int32_t * n_suppress_tokens) {
+    const std::vector<llama_token> & tokens = vocab->get_suppress_tokens();
+    if (n_suppress_tokens) {
+        *n_suppress_tokens = (int32_t) tokens.size();
+    }
+    return tokens.data();
+}
+
 llama_token llama_vocab_fim_pre(const struct llama_vocab * vocab) {
     return vocab->token_fim_pre();
 }
diff --git a/src/models/gemma4.cpp b/src/models/gemma4.cpp
index 6a96979ce..e44f423bd 100644
--- a/src/models/gemma4.cpp
+++ b/src/models/gemma4.cpp
@@ -142,33 +142,6 @@ static ggml_tensor * ggml_view_2d_slice(ggml_context * ctx0, ggml_tensor * x, in
                         idx * x->ne[0] * x->ne[1] * ggml_element_size(x));
 }
 
-// TODO @ngxson : maybe improve this in the future
-class llm_graph_input_logits_bias : public llm_graph_input_i {
-public:
-    llm_graph_input_logits_bias(const llama_vocab & vocab) {
-        arr.resize(vocab.n_tokens(), 0.0f);
-        for (llama_token id : vocab.get_suppress_tokens()) {
-            if (0 <= id && id < (int32_t)vocab.n_tokens()) {
-                arr[id] = -INFINITY;
-            }
-        }
-    }
-    virtual ~llm_graph_input_logits_bias() = default;
-
-    void set_input(const llama_ubatch * /*ubatch*/) override {
-        const int64_t n_vocab = arr.size();
-        ggml_backend_tensor_set(logits_bias, arr.data(), 0, n_vocab*ggml_element_size(logits_bias));
-    }
-
-    bool can_reuse(const llm_graph_params & /*params*/) override {
-        return true;
-    }
-
-    ggml_tensor * logits_bias = nullptr; // F32 [n_vocab]
-
-    std::vector<float> arr;
-};
-
 llama_model_gemma4::graph::graph(const llama_model & model, const llm_graph_params & params) :
         llm_graph_context(params),
         model(model),
@@ -429,16 +402,6 @@ llama_model_gemma4::graph::graph(const llama_model & model, const llm_graph_para
         cur = ggml_scale(ctx0, cur, hparams.f_final_logit_softcapping);
     }
 
-    // apply logits bias if needed (e.g. for gemma4_unified patch)
-    // this is to mirror the suppress_tokens patch on transformers, to avoid model from outputing <image|> and <audio|> tokens (which is a known issue related to the checkpoint)
-    // TODO: maybe handle this inside the sampling system in the future
-    if (!model.vocab.get_suppress_tokens().empty()) {
-        auto inp_bias = std::make_unique<llm_graph_input_logits_bias>(model.vocab);
-        inp_bias->logits_bias = ggml_new_tensor_1d(ctx0, GGML_TYPE_F32, inp_bias->arr.size());
-        cur = ggml_add(ctx0, cur, inp_bias->logits_bias);
-        res->add_input(std::move(inp_bias));
-    }
-
     cb(cur, "result_output", -1);
     res->t_logits = cur;
 

From 3018a11e79e489b657dbb77c95694889ccff92df Mon Sep 17 00:00:00 2001
From: Aleksander Grygier <aleksander.grygier@gmail.com>
Date: Wed, 29 Jul 2026 19:25:13 +0200
Subject: [PATCH 03/30] fix: increase greeting spacing on md screens (#26287)

---
 .../components/app/chat/ChatScreen/ChatScreenGreeting.svelte    | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/ui/src/lib/components/app/chat/ChatScreen/ChatScreenGreeting.svelte b/tools/ui/src/lib/components/app/chat/ChatScreen/ChatScreenGreeting.svelte
index 018949aff..5b44bcf85 100644
--- a/tools/ui/src/lib/components/app/chat/ChatScreen/ChatScreenGreeting.svelte
+++ b/tools/ui/src/lib/components/app/chat/ChatScreen/ChatScreenGreeting.svelte
@@ -11,7 +11,7 @@
 <div
 	class={[
 		'pointer-events-none mb-4 hidden px-4 text-center text-balance',
-		isEmpty && 'mb-[calc(50dvh-8rem)] md:mb-6 pointer-events-auto block!'
+		isEmpty && 'mb-[calc(50dvh-8rem)] md:mb-8 pointer-events-auto block!'
 	]}
 >
 	<h1 class="mb-2 text-2xl font-semibold tracking-tight md:text-3xl">Hello there</h1>

From 64d528be72d083fcaa76f3224ee88c742c2dd975 Mon Sep 17 00:00:00 2001
From: Tunahan <115956684+tnhnyzc@users.noreply.github.com>
Date: Thu, 30 Jul 2026 05:55:58 +0200
Subject: [PATCH 04/30] mimo2: address MTP review feedback (#26228)

Co-authored-by: tnhnyc <115956684+tnhnyc@users.noreply.github.com>
---
 gguf-py/gguf/constants.py |   3 +
 src/llama-model.cpp       |   4 +-
 src/models/mimo2.cpp      | 202 ++++++++++++++++++++++++++++++++++----
 src/models/models.h       |   4 +
 4 files changed, 191 insertions(+), 22 deletions(-)

diff --git a/gguf-py/gguf/constants.py b/gguf-py/gguf/constants.py
index 124ea28b0..650f1c8a5 100644
--- a/gguf-py/gguf/constants.py
+++ b/gguf-py/gguf/constants.py
@@ -4440,8 +4440,11 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
         MODEL_TENSOR.FFN_EXP_PROBS_B,
         MODEL_TENSOR.LAYER_OUT_NORM,
         MODEL_TENSOR.NEXTN_EH_PROJ,
+        MODEL_TENSOR.NEXTN_EMBED_TOKENS,
         MODEL_TENSOR.NEXTN_ENORM,
         MODEL_TENSOR.NEXTN_HNORM,
+        MODEL_TENSOR.NEXTN_SHARED_HEAD_HEAD,
+        MODEL_TENSOR.NEXTN_SHARED_HEAD_NORM,
     ],
     MODEL_ARCH.STEP35: [
         MODEL_TENSOR.TOKEN_EMBD,
diff --git a/src/llama-model.cpp b/src/llama-model.cpp
index a8422ff60..5e501f983 100644
--- a/src/llama-model.cpp
+++ b/src/llama-model.cpp
@@ -2243,7 +2243,9 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
                         filter = [&](uint32_t il) { return il >= hparams.n_layer(); };
                     }
 
-                    if ((arch == LLM_ARCH_STEP35 || arch == LLM_ARCH_HY_V3 || arch == LLM_ARCH_GLM_DSA) && hparams.n_layer_nextn > 0) {
+                    if ((arch == LLM_ARCH_STEP35 || arch == LLM_ARCH_HY_V3 || arch == LLM_ARCH_GLM_DSA ||
+                            arch == LLM_ARCH_MIMO2) &&
+                            hparams.n_layer_nextn > 0) {
                         if (params.ctx_type == LLAMA_CONTEXT_TYPE_MTP) {
                             filter = [&](uint32_t il) { return il >= hparams.n_layer(); };
                         } else {
diff --git a/src/models/mimo2.cpp b/src/models/mimo2.cpp
index 889891605..4080a934c 100644
--- a/src/models/mimo2.cpp
+++ b/src/models/mimo2.cpp
@@ -25,9 +25,13 @@ void llama_model_mimo2::load_arch_hparams(llama_model_loader & ml) {
     }
 }
 
-void llama_model_mimo2::load_arch_tensors(llama_model_loader &) {
+void llama_model_mimo2::load_arch_tensors(llama_model_loader & ml) {
     LLAMA_LOAD_LOCALS;
 
+    const std::string mtp_probe = "blk." + std::to_string(n_layer) + ".nextn.eh_proj.weight";
+    const bool trunk_only = (hparams.n_layer_nextn > 0) && (ml.get_weight(mtp_probe.c_str()) == nullptr);
+    const int  mtp_flags  = trunk_only ? TENSOR_NOT_REQUIRED : 0;
+
     tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 0);
 
     // output
@@ -40,41 +44,46 @@ void llama_model_mimo2::load_arch_tensors(llama_model_loader &) {
         uint32_t n_embd_v_gqa = hparams.n_embd_v_gqa(i);
         uint32_t n_head = hparams.n_head(i);
 
-        // NextN/MTP layers (the last n_nextn blocks) are preserved but disabled pending support
         const bool is_nextn = i >= n_layer;
-        const int  skip     = is_nextn ? TENSOR_SKIP : 0;
+        const int  flags    = is_nextn ? mtp_flags : 0;
 
-        create_tensor_qkv(layer, i, n_embd, n_embd_head_k * n_head, n_embd_k_gqa, n_embd_v_gqa, skip);
-        layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), { n_embd_head_v * n_head, n_embd }, skip);
+        create_tensor_qkv(layer, i, n_embd, n_embd_head_k * n_head, n_embd_k_gqa, n_embd_v_gqa, flags);
+        layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), { n_embd_head_v * n_head, n_embd }, flags);
 
-        layer.attn_norm  = create_tensor(tn(LLM_TENSOR_ATTN_NORM,  "weight", i), {n_embd}, skip);
-        layer.attn_sinks = create_tensor(tn(LLM_TENSOR_ATTN_SINKS, "weight", i), {n_head}, TENSOR_NOT_REQUIRED | skip);
+        layer.attn_norm  = create_tensor(tn(LLM_TENSOR_ATTN_NORM,  "weight", i), {n_embd}, flags);
+        layer.attn_sinks = create_tensor(tn(LLM_TENSOR_ATTN_SINKS, "weight", i), {n_head}, TENSOR_NOT_REQUIRED | flags);
 
-        layer.ffn_norm = create_tensor(tn(LLM_TENSOR_FFN_NORM, "weight", i), {n_embd}, skip);
+        layer.ffn_norm = create_tensor(tn(LLM_TENSOR_FFN_NORM, "weight", i), {n_embd}, flags);
 
         // non-MoE branch
-        layer.ffn_gate = create_tensor(tn(LLM_TENSOR_FFN_GATE, "weight", i), {n_embd,   n_ff}, TENSOR_NOT_REQUIRED | skip);
-        layer.ffn_down = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), {  n_ff, n_embd}, TENSOR_NOT_REQUIRED | skip);
-        layer.ffn_up   = create_tensor(tn(LLM_TENSOR_FFN_UP,   "weight", i), {n_embd,   n_ff}, TENSOR_NOT_REQUIRED | skip);
+        layer.ffn_gate = create_tensor(tn(LLM_TENSOR_FFN_GATE, "weight", i), {n_embd,   n_ff}, TENSOR_NOT_REQUIRED | flags);
+        layer.ffn_down = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), {  n_ff, n_embd}, TENSOR_NOT_REQUIRED | flags);
+        layer.ffn_up   = create_tensor(tn(LLM_TENSOR_FFN_UP,   "weight", i), {n_embd,   n_ff}, TENSOR_NOT_REQUIRED | flags);
 
         // MoE branch
         int64_t n_ff_exp = hparams.n_ff_exp;
-        layer.ffn_gate_inp  = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP,  "weight", i), {n_embd, n_expert}, TENSOR_NOT_REQUIRED | skip);
-        layer.ffn_gate_exps = create_tensor(tn(LLM_TENSOR_FFN_GATE_EXPS, "weight", i), {n_embd, n_ff_exp,   n_expert}, TENSOR_NOT_REQUIRED | skip);
-        layer.ffn_down_exps = create_tensor(tn(LLM_TENSOR_FFN_DOWN_EXPS, "weight", i), {n_ff_exp,   n_embd, n_expert}, TENSOR_NOT_REQUIRED | skip);
-        layer.ffn_up_exps   = create_tensor(tn(LLM_TENSOR_FFN_UP_EXPS,   "weight", i), {n_embd, n_ff_exp,   n_expert}, TENSOR_NOT_REQUIRED | skip);
-        layer.ffn_exp_probs_b = create_tensor(tn(LLM_TENSOR_FFN_EXP_PROBS_B, "bias", i), {n_expert}, TENSOR_NOT_REQUIRED | skip);
+        layer.ffn_gate_inp  = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP,  "weight", i), {n_embd, n_expert}, TENSOR_NOT_REQUIRED | flags);
+        layer.ffn_gate_exps = create_tensor(tn(LLM_TENSOR_FFN_GATE_EXPS, "weight", i), {n_embd, n_ff_exp,   n_expert}, TENSOR_NOT_REQUIRED | flags);
+        layer.ffn_down_exps = create_tensor(tn(LLM_TENSOR_FFN_DOWN_EXPS, "weight", i), {n_ff_exp,   n_embd, n_expert}, TENSOR_NOT_REQUIRED | flags);
+        layer.ffn_up_exps   = create_tensor(tn(LLM_TENSOR_FFN_UP_EXPS,   "weight", i), {n_embd, n_ff_exp,   n_expert}, TENSOR_NOT_REQUIRED | flags);
+        layer.ffn_exp_probs_b = create_tensor(tn(LLM_TENSOR_FFN_EXP_PROBS_B, "bias", i), {n_expert}, TENSOR_NOT_REQUIRED | flags);
 
         if (is_nextn) {
-            layer.nextn.eh_proj  = create_tensor(tn(LLM_TENSOR_NEXTN_EH_PROJ, "weight", i), {2 * n_embd, n_embd}, skip);
-            layer.nextn.enorm    = create_tensor(tn(LLM_TENSOR_NEXTN_ENORM,   "weight", i), {n_embd}, skip);
-            layer.nextn.hnorm    = create_tensor(tn(LLM_TENSOR_NEXTN_HNORM,   "weight", i), {n_embd}, skip);
-            layer.layer_out_norm = create_tensor(tn(LLM_TENSOR_LAYER_OUT_NORM, "weight", i), {n_embd}, skip);
+            layer.nextn.eh_proj          = create_tensor(tn(LLM_TENSOR_NEXTN_EH_PROJ,          "weight", i), {2 * n_embd, n_embd}, flags);
+            layer.nextn.enorm            = create_tensor(tn(LLM_TENSOR_NEXTN_ENORM,            "weight", i), {n_embd}, flags);
+            layer.nextn.hnorm            = create_tensor(tn(LLM_TENSOR_NEXTN_HNORM,            "weight", i), {n_embd}, flags);
+            layer.nextn.embed_tokens     = create_tensor(tn(LLM_TENSOR_NEXTN_EMBED_TOKENS,     "weight", i), {n_embd, n_vocab}, TENSOR_NOT_REQUIRED | flags);
+            layer.nextn.shared_head_head = create_tensor(tn(LLM_TENSOR_NEXTN_SHARED_HEAD_HEAD, "weight", i), {n_embd, n_vocab}, TENSOR_NOT_REQUIRED | flags);
+            layer.nextn.shared_head_norm = create_tensor(tn(LLM_TENSOR_NEXTN_SHARED_HEAD_NORM, "weight", i), {n_embd}, TENSOR_NOT_REQUIRED | flags);
+            layer.layer_out_norm         = create_tensor(tn(LLM_TENSOR_LAYER_OUT_NORM,         "weight", i), {n_embd}, TENSOR_NOT_REQUIRED | flags);
         }
     }
 }
 
 std::unique_ptr<llm_graph_context> llama_model_mimo2::build_arch_graph(const llm_graph_params & params) const {
+    if (params.gtype == LLM_GRAPH_TYPE_DECODER_MTP) {
+        return std::make_unique<graph_mtp>(*this, params);
+    }
     return std::make_unique<graph>(*this, params);
 }
 
@@ -89,6 +98,8 @@ llama_model_mimo2::graph::graph(const llama_model & model, const llm_graph_param
     ggml_tensor * inp_out_ids = build_inp_out_ids();
 
     const float v_scale = hparams.f_attn_value_scale;
+    const bool emit_h_nextn = cparams.embeddings_nextn;
+    const bool crop_last_layer = inp_out_ids && (!emit_h_nextn || cparams.embeddings_nextn_masked);
 
     for (int il = 0; il < n_layer; ++il) {
         ggml_tensor * inpSA = inpL;
@@ -168,7 +179,7 @@ llama_model_mimo2::graph::graph(const llama_model & model, const llm_graph_param
             }
         }
 
-        if (il == n_layer - 1 && inp_out_ids) {
+        if (il == n_layer - 1 && crop_last_layer) {
             cur   = ggml_get_rows(ctx0,   cur, inp_out_ids);
             inpSA = ggml_get_rows(ctx0, inpSA, inp_out_ids);
         }
@@ -218,6 +229,15 @@ llama_model_mimo2::graph::graph(const llama_model & model, const llm_graph_param
 
     cur = inpL;
 
+    if (emit_h_nextn) {
+        cb(cur, "h_nextn", -1);
+        res->t_h_nextn = cur;
+
+        if (!cparams.embeddings_nextn_masked && inp_out_ids) {
+            cur = ggml_get_rows(ctx0, cur, inp_out_ids);
+        }
+    }
+
     cur = build_norm(cur,
             model.output_norm, NULL,
             LLM_NORM_RMS, -1);
@@ -233,3 +253,143 @@ llama_model_mimo2::graph::graph(const llama_model & model, const llm_graph_param
 
     ggml_build_forward_expand(gf, cur);
 }
+
+// Mirrors MiMo's appended NextN block: normalize and fuse token and hidden inputs, run the decoder block,
+// expose its pre-head-norm state to the next draft step, then apply the shared output norm and LM head.
+// Converted checkpoints may store that shared norm as layer_out_norm, so it remains in the fallback chain.
+llama_model_mimo2::graph_mtp::graph_mtp(const llama_model & model, const llm_graph_params & params)
+    : llm_graph_context(params) {
+    GGML_ASSERT(hparams.n_layer_nextn > 0 && "MIMO2 MTP requires n_layer_nextn > 0");
+
+    const int il = hparams.n_layer() + cparams.nextn_layer_offset;
+    GGML_ASSERT(cparams.nextn_layer_offset >= 0 &&
+                cparams.nextn_layer_offset < (int) hparams.n_layer_nextn &&
+                "nextn_layer_offset out of range [0, n_layer_nextn)");
+
+    const auto & layer = model.layers[il];
+    GGML_ASSERT(layer.nextn.eh_proj && "MIMO2 MTP block missing nextn.eh_proj");
+    GGML_ASSERT(layer.nextn.enorm   && "MIMO2 MTP block missing nextn.enorm");
+    GGML_ASSERT(layer.nextn.hnorm   && "MIMO2 MTP block missing nextn.hnorm");
+    GGML_ASSERT(layer.wqkv          && "MIMO2 MTP requires fused attn_qkv");
+
+    const uint32_t n_head_l    = hparams.n_head(il);
+    const uint32_t n_head_kv_l = hparams.n_head_kv(il);
+
+    const float freq_base_l  = model.get_rope_freq_base(cparams, il);
+    const float freq_scale_l = model.get_rope_freq_scale(cparams, il);
+    const float v_scale      = hparams.f_attn_value_scale;
+
+    auto inp = std::make_unique<llm_graph_input_embd>(hparams.n_embd);
+
+    inp->tokens = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, n_tokens);
+    ggml_set_input(inp->tokens);
+
+    inp->embd = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, hparams.n_embd, n_tokens);
+    ggml_set_input(inp->embd);
+    ggml_set_name(inp->embd, "mtp_h_input");
+
+    ggml_tensor * tok_embd_w = layer.nextn.embed_tokens ? layer.nextn.embed_tokens : model.tok_embd;
+    ggml_tensor * h_input    = inp->embd;
+    ggml_tensor * tok_embd   = ggml_get_rows(ctx0, tok_embd_w, inp->tokens);
+    cb(tok_embd, "mtp_tok_embd", il);
+
+    res->add_input(std::move(inp));
+
+    ggml_tensor * inp_pos     = build_inp_pos();
+    ggml_tensor * inp_out_ids = build_inp_out_ids();
+    auto        * inp_attn    = build_attn_inp_kv_iswa();
+
+    ggml_tensor * h_norm = build_norm(h_input, layer.nextn.hnorm, nullptr, LLM_NORM_RMS, il);
+    cb(h_norm, "mtp_hnorm", il);
+
+    ggml_tensor * e_norm = build_norm(tok_embd, layer.nextn.enorm, nullptr, LLM_NORM_RMS, il);
+    cb(e_norm, "mtp_enorm", il);
+
+    ggml_tensor * concat = ggml_concat(ctx0, e_norm, h_norm, /*dim=*/ 0);
+    cb(concat, "mtp_concat", il);
+
+    ggml_tensor * cur = build_lora_mm(layer.nextn.eh_proj, concat, layer.nextn.eh_proj_s);
+    cb(cur, "mtp_eh_proj", il);
+
+    ggml_tensor * inpSA = cur;
+
+    cur = build_norm(cur, layer.attn_norm, nullptr, LLM_NORM_RMS, il);
+    cb(cur, "mtp_attn_norm", il);
+
+    ggml_tensor * qkv = build_lora_mm(layer.wqkv, cur, layer.wqkv_s);
+    cb(qkv, "mtp_wqkv", il);
+
+    const size_t row_k    = ggml_row_size(qkv->type, n_embd_head_k);
+    const size_t row_v    = ggml_row_size(qkv->type, n_embd_head_v);
+    const size_t row_full = qkv->nb[1];
+    const size_t k_off    = row_k * n_head_l;
+    const size_t v_off    = k_off + row_k * n_head_kv_l;
+
+    ggml_tensor * Qcur = ggml_view_3d(ctx0, qkv, n_embd_head_k, n_head_l,    n_tokens, row_k, row_full, 0);
+    ggml_tensor * Kcur = ggml_view_3d(ctx0, qkv, n_embd_head_k, n_head_kv_l, n_tokens, row_k, row_full, k_off);
+    ggml_tensor * Vcur = ggml_view_3d(ctx0, qkv, n_embd_head_v, n_head_kv_l, n_tokens, row_v, row_full, v_off);
+
+    Qcur = ggml_rope_ext(
+        ctx0, Qcur, inp_pos, nullptr,
+        n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
+        ext_factor, attn_factor, beta_fast, beta_slow);
+
+    Kcur = ggml_rope_ext(
+        ctx0, Kcur, inp_pos, nullptr,
+        n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
+        ext_factor, attn_factor, beta_fast, beta_slow);
+
+    cb(Qcur, "mtp_Qcur", il);
+    cb(Kcur, "mtp_Kcur", il);
+    cb(Vcur, "mtp_Vcur", il);
+
+    cur = build_attn(inp_attn,
+            layer.wo, nullptr, layer.wo_s,
+            Qcur, Kcur, Vcur, nullptr, layer.attn_sinks, nullptr,
+            1.0f / sqrtf(float(n_embd_head_k)), il);
+    cb(cur, "mtp_attn_out", il);
+
+    if (v_scale) {
+        cur = ggml_scale(ctx0, cur, v_scale);
+        cb(cur, "mtp_attn_out_scaled", il);
+    }
+
+    ggml_tensor * ffn_inp = ggml_add(ctx0, cur, inpSA);
+    cb(ffn_inp, "mtp_ffn_inp", il);
+
+    cur = build_norm(ffn_inp, layer.ffn_norm, nullptr, LLM_NORM_RMS, il);
+    cb(cur, "mtp_ffn_norm", il);
+
+    GGML_ASSERT(layer.ffn_gate && layer.ffn_down && layer.ffn_up && "MIMO2 MTP requires dense FFN tensors");
+    cur = build_ffn(cur,
+            layer.ffn_up,   layer.ffn_up_b,   nullptr,
+            layer.ffn_gate, layer.ffn_gate_b, nullptr,
+            layer.ffn_down, layer.ffn_down_b, nullptr,
+            nullptr,
+            LLM_FFN_SILU, LLM_FFN_PAR, il);
+    cb(cur, "mtp_ffn_out", il);
+
+    cur = ggml_add(ctx0, cur, ffn_inp);
+    cb(cur, "mtp_post_ffn", il);
+
+    cur = ggml_get_rows(ctx0, cur, inp_out_ids);
+
+    cb(cur, "h_nextn", -1);
+    res->t_h_nextn = cur;
+
+    ggml_tensor * head_norm_w = layer.nextn.shared_head_norm
+            ? layer.nextn.shared_head_norm
+            : (layer.layer_out_norm ? layer.layer_out_norm : model.output_norm);
+    GGML_ASSERT(head_norm_w && "MIMO2 MTP missing head norm fallback");
+    cur = build_norm(cur, head_norm_w, nullptr, LLM_NORM_RMS, -1);
+    cb(cur, "mtp_shared_head_norm", -1);
+
+    ggml_tensor * head_w = layer.nextn.shared_head_head ? layer.nextn.shared_head_head : model.output;
+    ggml_tensor * head_s = layer.nextn.shared_head_head ? layer.nextn.shared_head_head_s : model.output_s;
+    GGML_ASSERT(head_w && "MIMO2 MTP missing LM head fallback");
+    cur = build_lora_mm(head_w, cur, head_s);
+    cb(cur, "result_output", -1);
+
+    res->t_logits = cur;
+    ggml_build_forward_expand(gf, cur);
+}
diff --git a/src/models/models.h b/src/models/models.h
index c73136f3b..bb372ece8 100644
--- a/src/models/models.h
+++ b/src/models/models.h
@@ -2127,6 +2127,10 @@ struct llama_model_mimo2 : public llama_model_base {
         graph(const llama_model & model, const llm_graph_params & params);
     };
 
+    struct graph_mtp : public llm_graph_context {
+        graph_mtp(const llama_model & model, const llm_graph_params & params);
+    };
+
     std::unique_ptr<llm_graph_context> build_arch_graph(const llm_graph_params & params) const override;
 };
 

From a6a77bc48d05d8e1f5467b91ec0b5bcc8897b86c Mon Sep 17 00:00:00 2001
From: Neo Zhang <zhang.jianyu@outlook.com>
Date: Thu, 30 Jul 2026 14:04:58 +0800
Subject: [PATCH 05/30] [UT] enhance UT to show all real unsupported backends
 (#25234)

* enhance UT to show real unsupported backends

* cont : simplify

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
---
 tests/test-backend-ops.cpp | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/tests/test-backend-ops.cpp b/tests/test-backend-ops.cpp
index a5b660f47..d0eb173a8 100644
--- a/tests/test-backend-ops.cpp
+++ b/tests/test-backend-ops.cpp
@@ -1350,18 +1350,22 @@ struct test_case {
 
         // check if the backends support the ops
         bool supported = true;
+        std::string unsupported_str;
         for (ggml_backend_t backend : {backend1, backend2}) {
             for (ggml_tensor * t = ggml_get_first_tensor(ctx.get()); t != NULL; t = ggml_get_next_tensor(ctx.get(), t)) {
                 if (!ggml_backend_supports_op(backend, t)) {
                     supported = false;
-                    break;
+                    if (unsupported_str.empty()) {
+                        unsupported_str = std::string(ggml_backend_name(backend));
+                    } else {
+                        unsupported_str += ", " + std::string(ggml_backend_name(backend));
+                    }
                 }
             }
         }
 
         if (!supported) {
-            // Create test result for unsupported operation
-            test_result result(ggml_backend_name(backend1), current_op_name, vars(), "test",
+            test_result result(unsupported_str, current_op_name, vars(), "test",
                              false, false, "not supported");
 
             print_test_result_locked(output_printer, result);

From 32703b42d6d8a74336bf27613705cbc76e2363ee Mon Sep 17 00:00:00 2001
From: Jonathan Clohessy <jonathan.clohessy@arm.com>
Date: Thu, 30 Jul 2026 07:17:30 +0100
Subject: [PATCH 06/30] ggml : Fix issue with kleidiai ci and stringop overflow
 warning (#26277)

Signed-off-by: Jonathan Clohessy <Jonathan.Clohessy@arm.com>
---
 ggml/src/ggml-cpu/repack.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ggml/src/ggml-cpu/repack.cpp b/ggml/src/ggml-cpu/repack.cpp
index f18758f16..9689ca3ce 100644
--- a/ggml/src/ggml-cpu/repack.cpp
+++ b/ggml/src/ggml-cpu/repack.cpp
@@ -2739,7 +2739,7 @@ static block_q8_0x4 make_block_q8_0x4(block_q8_0 * in, unsigned int blck_size_in
     return out;
 }
 
-static block_q4_0x4 make_block_q4_0x4(block_q4_0 * in, unsigned int blck_size_interleave) {
+static block_q4_0x4 make_block_q4_0x4(block_q4_0 * in, int blck_size_interleave) {
     block_q4_0x4 out;
 
     for (int i = 0; i < 4; i++) {

From 21a5f5b7f919834e37fd492158aa24dc2f5e3b26 Mon Sep 17 00:00:00 2001
From: Aleksander Grygier <aleksander.grygier@gmail.com>
Date: Thu, 30 Jul 2026 10:10:37 +0200
Subject: [PATCH 07/30] ui: IndexedDB and Conversations data fixes (#26278)

* fix: single-flight conversations store init

* refactor: remove unused legacy-migration util

* fix: make createSystemMessage transactional

* fix: delete message branches cascading on edit/regenerate

* fix: stop stamping lastModified on conversation metadata updates

* fix: count cascaded forks in bulk delete toast, bulkify deleteAll

* refactor: drop redundant conversation list respreads

* refactor: create conversation in a single write

* fix: use table constant in toggleConversationPin

* fix: keep the system message placeholder out of the edit form

* fix: keep focus in the system message editor after opening it

* fix: focus the main chat form after submitting a system message

* fix: update timestamp of the correct conversation on stream completion
---
 .../ChatFormActionAddDropdown.svelte          |  19 +-
 .../ChatMessage/ChatMessage.svelte            |  19 +-
 .../app/chat/ChatScreen/ChatScreenForm.svelte |  14 +-
 tools/ui/src/lib/services/database.service.ts |  57 +--
 tools/ui/src/lib/stores/chat.svelte.ts        |   7 +-
 .../ui/src/lib/stores/conversations.svelte.ts |  76 ++--
 tools/ui/src/lib/utils/legacy-migration.ts    | 364 ------------------
 7 files changed, 126 insertions(+), 430 deletions(-)
 delete mode 100644 tools/ui/src/lib/utils/legacy-migration.ts

diff --git a/tools/ui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionAdd/ChatFormActionAddDropdown.svelte b/tools/ui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionAdd/ChatFormActionAddDropdown.svelte
index 905c2fe6f..f81dcf09c 100644
--- a/tools/ui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionAdd/ChatFormActionAddDropdown.svelte
+++ b/tools/ui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionAdd/ChatFormActionAddDropdown.svelte
@@ -48,6 +48,9 @@
 	}: Props = $props();
 
 	let dropdownOpen = $state(false);
+	// The system message action moves focus to the message editor, so the menu
+	// must not restore focus to the trigger on close
+	let suppressCloseAutoFocus = false;
 
 	function handleMcpSettingsClick() {
 		dropdownOpen = false;
@@ -96,7 +99,16 @@
 			</Tooltip.Content>
 		</Tooltip.Root>
 
-		<DropdownMenu.Content align="start" class="w-52">
+		<DropdownMenu.Content
+			align="start"
+			class="w-52"
+			onCloseAutoFocus={(e) => {
+				if (suppressCloseAutoFocus) {
+					suppressCloseAutoFocus = false;
+					e.preventDefault();
+				}
+			}}
+		>
 			<ChatFormActionAddReasoningSubmenu />
 
 			<DropdownMenu.Separator />
@@ -148,7 +160,10 @@
 
 			<DropdownMenu.Item
 				class="flex cursor-pointer items-center gap-2"
-				onclick={onSystemPromptClick}
+				onclick={() => {
+					suppressCloseAutoFocus = true;
+					onSystemPromptClick?.();
+				}}
 			>
 				<MessageSquare class={ICON_CLASS_DEFAULT} />
 
diff --git a/tools/ui/src/lib/components/app/chat/ChatMessages/ChatMessage/ChatMessage.svelte b/tools/ui/src/lib/components/app/chat/ChatMessages/ChatMessage/ChatMessage.svelte
index 8e8a14ac3..b8068f790 100644
--- a/tools/ui/src/lib/components/app/chat/ChatMessages/ChatMessage/ChatMessage.svelte
+++ b/tools/ui/src/lib/components/app/chat/ChatMessages/ChatMessage/ChatMessage.svelte
@@ -2,6 +2,7 @@
 	import { goto } from '$app/navigation';
 	import { getChatActionsContext, setMessageEditContext } from '$lib/contexts';
 	import { chatStore, pendingEditMessageId } from '$lib/stores/chat.svelte';
+	import { isMobile } from '$lib/stores/viewport.svelte';
 	import { conversationsStore } from '$lib/stores/conversations.svelte';
 	import { DatabaseService } from '$lib/services/database.service';
 	import { SYSTEM_MESSAGE_PLACEHOLDER } from '$lib/constants';
@@ -46,7 +47,14 @@
 		assistantMessages: number;
 		messageTypes: string[];
 	} | null>(null);
-	let editedContent = $derived(message.content);
+	// The system message placeholder must never surface as editable content; keeping
+	// it in the derived (not just in handleEdit) guards against prop invalidation
+	// reverting the override while editing
+	let editedContent = $derived(
+		message.role === MessageRole.SYSTEM && message.content === SYSTEM_MESSAGE_PLACEHOLDER
+			? ''
+			: message.content
+	);
 
 	let rawEditContent = $derived.by(() => {
 		if (message.role !== MessageRole.ASSISTANT) return undefined;
@@ -265,6 +273,12 @@
 		chatActions.navigateToSibling(siblingId);
 	}
 
+	// After the system message flow ends, hand focus to the main chat form
+	function focusMainChatForm() {
+		if (isMobile.current) return;
+		document.querySelector<HTMLTextAreaElement>('.chat-screen-form-wrapper textarea')?.focus();
+	}
+
 	async function handleSaveEdit() {
 		if (message.role === MessageRole.SYSTEM) {
 			// System messages: update in place without branching
@@ -276,6 +290,8 @@
 				isEditing = false;
 				if (conversationDeleted) {
 					goto(ROUTES.START);
+				} else {
+					focusMainChatForm();
 				}
 				return;
 			}
@@ -285,6 +301,7 @@
 			if (index !== -1) {
 				conversationsStore.updateMessageAtIndex(index, { content: newContent });
 			}
+			focusMainChatForm();
 		} else if (message.role === MessageRole.USER) {
 			const finalExtras = await getMergedExtras();
 			chatActions.editWithBranching(message, editedContent.trim(), finalExtras);
diff --git a/tools/ui/src/lib/components/app/chat/ChatScreen/ChatScreenForm.svelte b/tools/ui/src/lib/components/app/chat/ChatScreen/ChatScreenForm.svelte
index 600180742..8eb17eeae 100644
--- a/tools/ui/src/lib/components/app/chat/ChatScreen/ChatScreenForm.svelte
+++ b/tools/ui/src/lib/components/app/chat/ChatScreen/ChatScreenForm.svelte
@@ -106,15 +106,23 @@
 		onFileRemove?.(fileId);
 	}
 
+	// Auto-focus must not steal focus already claimed elsewhere (e.g. the system
+	// message editor opened just before a navigation)
+	function focusFormUnlessCaptured() {
+		const active = document.activeElement;
+		if (active instanceof HTMLTextAreaElement || active instanceof HTMLInputElement) return;
+		chatFormRef?.focus();
+	}
+
 	onMount(() => {
 		if (!isMobile.current) {
-			setTimeout(() => chatFormRef?.focus(), 100);
+			setTimeout(focusFormUnlessCaptured, 100);
 		}
 	});
 
 	afterNavigate((navigation) => {
 		if (navigation?.from != null && !isMobile.current) {
-			setTimeout(() => chatFormRef?.focus(), 100);
+			setTimeout(focusFormUnlessCaptured, 100);
 		}
 	});
 
@@ -127,7 +135,7 @@
 
 	$effect(() => {
 		if (previousIsLoading && !isLoading) {
-			setTimeout(() => chatFormRef?.focus(), 10);
+			setTimeout(focusFormUnlessCaptured, 10);
 		}
 
 		previousIsLoading = isLoading;
diff --git a/tools/ui/src/lib/services/database.service.ts b/tools/ui/src/lib/services/database.service.ts
index bc65caaca..0a7c59b9c 100644
--- a/tools/ui/src/lib/services/database.service.ts
+++ b/tools/ui/src/lib/services/database.service.ts
@@ -31,14 +31,19 @@ export class DatabaseService {
 	 * Creates a new conversation.
 	 *
 	 * @param name - Name of the conversation
+	 * @param fields - Optional extra fields (e.g. reasoningEffort)
 	 * @returns The created conversation
 	 */
-	static async createConversation(name: string): Promise<DatabaseConversation> {
+	static async createConversation(
+		name: string,
+		fields?: Partial<Omit<DatabaseConversation, 'id' | 'name' | 'lastModified'>>
+	): Promise<DatabaseConversation> {
 		const conversation: DatabaseConversation = {
 			id: uuid(),
 			name,
 			lastModified: Date.now(),
-			currNode: ''
+			currNode: '',
+			...fields
 		};
 
 		await db[IDXDB_TABLES.conversations].add(conversation);
@@ -137,7 +142,7 @@ export class DatabaseService {
 	 * @param systemPrompt - The system prompt content (must be non-empty)
 	 * @param parentId - Parent message ID (typically the root message)
 	 * @returns The created system message
-	 * @throws Error if systemPrompt is empty
+	 * @throws Error if systemPrompt is empty or the parent message does not exist
 	 */
 	static async createSystemMessage(
 		convId: string,
@@ -149,27 +154,30 @@ export class DatabaseService {
 			throw new Error('Cannot create system message with empty content');
 		}
 
-		const systemMessage: DatabaseMessage = {
-			id: uuid(),
-			convId,
-			type: MessageRole.SYSTEM,
-			timestamp: Date.now(),
-			role: MessageRole.SYSTEM,
-			content: trimmedPrompt,
-			parent: parentId,
-			children: []
-		};
+		return await db.transaction('rw', db[IDXDB_TABLES.messages], async () => {
+			const parentMessage = await db[IDXDB_TABLES.messages].get(parentId);
+			if (!parentMessage) {
+				throw new Error(`Parent message ${parentId} not found`);
+			}
 
-		await db[IDXDB_TABLES.messages].add(systemMessage);
+			const systemMessage: DatabaseMessage = {
+				id: uuid(),
+				convId,
+				type: MessageRole.SYSTEM,
+				timestamp: Date.now(),
+				role: MessageRole.SYSTEM,
+				content: trimmedPrompt,
+				parent: parentId,
+				children: []
+			};
 
-		const parentMessage = await db[IDXDB_TABLES.messages].get(parentId);
-		if (parentMessage) {
+			await db[IDXDB_TABLES.messages].add(systemMessage);
 			await db[IDXDB_TABLES.messages].update(parentId, {
 				children: [...parentMessage.children, systemMessage.id]
 			});
-		}
 
-		return systemMessage;
+			return systemMessage;
+		});
 	}
 
 	/**
@@ -442,7 +450,8 @@ export class DatabaseService {
 	}
 
 	/**
-	 * Updates a conversation.
+	 * Updates a conversation. `lastModified` is never stamped implicitly;
+	 * pass it in `updates` to bump the conversation in recency ordering.
 	 *
 	 * @param id - Conversation ID
 	 * @param updates - Partial updates to apply
@@ -452,10 +461,7 @@ export class DatabaseService {
 		id: string,
 		updates: Partial<Omit<DatabaseConversation, 'id'>>
 	): Promise<void> {
-		await db[IDXDB_TABLES.conversations].update(id, {
-			...updates,
-			lastModified: Date.now()
-		});
+		await db[IDXDB_TABLES.conversations].update(id, updates);
 	}
 
 	/**
@@ -473,7 +479,7 @@ export class DatabaseService {
 	 * @returns The new pinned status
 	 */
 	static async toggleConversationPin(id: string): Promise<boolean> {
-		const conversation = await db.conversations.get(id);
+		const conversation = await db[IDXDB_TABLES.conversations].get(id);
 		if (!conversation) {
 			throw new Error(`Conversation ${id} not found`);
 		}
@@ -497,7 +503,6 @@ export class DatabaseService {
 		const result = new Map<string, boolean>();
 		if (cleanIds.length === 0) return result;
 
-		const now = Date.now();
 		await db.transaction('rw', db[IDXDB_TABLES.conversations], async () => {
 			const convs = await db[IDXDB_TABLES.conversations].bulkGet(cleanIds);
 			const updates: DatabaseConversation[] = [];
@@ -505,7 +510,7 @@ export class DatabaseService {
 				const conv = convs[i];
 				if (!conv) continue;
 				const newPinned = !conv.pinned;
-				updates.push({ ...conv, pinned: newPinned, lastModified: now });
+				updates.push({ ...conv, pinned: newPinned });
 				result.set(cleanIds[i], newPinned);
 			}
 			if (updates.length === 0) return;
diff --git a/tools/ui/src/lib/stores/chat.svelte.ts b/tools/ui/src/lib/stores/chat.svelte.ts
index 222723ab1..935ed8e16 100644
--- a/tools/ui/src/lib/stores/chat.svelte.ts
+++ b/tools/ui/src/lib/stores/chat.svelte.ts
@@ -1658,7 +1658,8 @@ class ChatStore {
 					generateConversationTitle(newContent, Boolean(config().titleGenerationUseFirstLine))
 				);
 			const messagesToRemove = conversationsStore.activeMessages.slice(messageIndex + 1);
-			for (const message of messagesToRemove) await DatabaseService.deleteMessage(message.id);
+			if (messagesToRemove.length > 0)
+				await DatabaseService.deleteMessageCascading(activeConv.id, messagesToRemove[0].id);
 			conversationsStore.sliceActiveMessages(messageIndex + 1);
 			conversationsStore.updateConversationTimestamp();
 			this.setChatLoading(activeConv.id, true);
@@ -1690,7 +1691,7 @@ class ChatStore {
 		const { index: messageIndex } = result;
 		try {
 			const messagesToRemove = conversationsStore.activeMessages.slice(messageIndex);
-			for (const message of messagesToRemove) await DatabaseService.deleteMessage(message.id);
+			await DatabaseService.deleteMessageCascading(activeConv.id, messagesToRemove[0].id);
 			conversationsStore.sliceActiveMessages(messageIndex);
 			conversationsStore.updateConversationTimestamp();
 			this.setChatLoading(activeConv.id, true);
@@ -2037,7 +2038,7 @@ class ChatStore {
 							timings
 						});
 
-						conversationsStore.updateConversationTimestamp();
+						conversationsStore.updateConversationTimestamp(msg.convId);
 
 						this.setChatLoading(msg.convId, false);
 						this.clearChatStreaming(msg.convId);
diff --git a/tools/ui/src/lib/stores/conversations.svelte.ts b/tools/ui/src/lib/stores/conversations.svelte.ts
index e467c8fad..1a1157c72 100644
--- a/tools/ui/src/lib/stores/conversations.svelte.ts
+++ b/tools/ui/src/lib/stores/conversations.svelte.ts
@@ -111,6 +111,9 @@ class ConversationsStore {
 		| ((messageId: string, updates: Partial<DatabaseMessage>) => void)
 		| null = null;
 
+	/** In-flight init run; shared by concurrent callers, reset on failure to allow retry */
+	private initPromise: Promise<void> | null = null;
+
 	/**
 	 *
 	 *
@@ -121,19 +124,25 @@ class ConversationsStore {
 
 	/**
 	 * Initialize the store by loading conversations from database.
-	 * Must be called once after app startup.
+	 * Safe to call multiple times: concurrent callers share a single run,
+	 * and a failed run can be retried by calling again.
 	 */
-	async init(): Promise<void> {
-		if (!browser) return;
-		if (this.isInitialized) return;
+	init(): Promise<void> {
+		if (!browser) return Promise.resolve();
+		if (this.initPromise) return this.initPromise;
 
-		try {
-			await MigrationService.runAllMigrations();
-			await this.loadConversations();
-			this.isInitialized = true;
-		} catch (error) {
-			console.error('Failed to initialize conversations:', error);
-		}
+		this.initPromise = (async () => {
+			try {
+				await MigrationService.runAllMigrations();
+				await this.loadConversations();
+				this.isInitialized = true;
+			} catch (error) {
+				console.error('Failed to initialize conversations:', error);
+				this.initPromise = null;
+			}
+		})();
+
+		return this.initPromise;
 	}
 
 	/**
@@ -237,15 +246,11 @@ class ConversationsStore {
 	 */
 	async createConversation(name?: string): Promise<string> {
 		const conversationName = name || `Chat ${new Date().toLocaleString()}`;
-		const conversation = await DatabaseService.createConversation(conversationName);
 
 		// No MCP override list is seeded: getAllMcpServerOverrides resolves
 		// servers without a per-conversation override to `mcpServers[i].enabled`,
 		// and only explicit toggles are stored on the conversation.
-
-		// Inherit the global reasoning default into the new conversation
-		conversation.reasoningEffort = this.pendingReasoningEffort;
-		await DatabaseService.updateConversation(conversation.id, {
+		const conversation = await DatabaseService.createConversation(conversationName, {
 			reasoningEffort: this.pendingReasoningEffort
 		});
 
@@ -358,10 +363,7 @@ class ConversationsStore {
 	async deleteAll(): Promise<void> {
 		try {
 			const allConversations = await DatabaseService.getAllConversations();
-
-			for (const conv of allConversations) {
-				await DatabaseService.deleteConversation(conv.id);
-			}
+			await DatabaseService.bulkDeleteConversations(allConversations.map((c) => c.id));
 
 			this.clearActiveConversation();
 			this.conversations = [];
@@ -412,7 +414,9 @@ class ConversationsStore {
 			}
 
 			toast.success(
-				convIds.length === 1 ? 'Conversation deleted' : `${convIds.length} conversations deleted`
+				idsToRemove.size === 1
+					? 'Conversation deleted'
+					: `${idsToRemove.size} conversations deleted`
 			);
 		} catch (error) {
 			console.error('Failed to bulk delete conversations:', error);
@@ -443,7 +447,6 @@ class ConversationsStore {
 				const newPinned = updates.get(this.conversations[i].id);
 				if (newPinned !== undefined) this.conversations[i].pinned = newPinned;
 			}
-			this.conversations = [...this.conversations];
 
 			toast.success(
 				convIds.length === 1
@@ -552,7 +555,6 @@ class ConversationsStore {
 
 			if (convIndex !== -1) {
 				this.conversations[convIndex].name = name;
-				this.conversations = [...this.conversations];
 			}
 
 			if (this.activeConversation?.id === convId) {
@@ -576,7 +578,6 @@ class ConversationsStore {
 
 			if (convIndex !== -1) {
 				this.conversations[convIndex].pinned = newPinnedState;
-				this.conversations = [...this.conversations];
 			}
 
 			if (this.activeConversation?.id === convId) {
@@ -591,18 +592,33 @@ class ConversationsStore {
 	}
 
 	/**
-	 * Updates conversation lastModified timestamp and moves it to top of list
+	 * Marks a conversation as recently active: stamps lastModified (persisted)
+	 * and moves it to the top of the list. Only message-activity flows call
+	 * this; metadata updates (rename, pin, settings) do not.
+	 *
+	 * @param convId - Conversation that produced the activity, defaults to the active one
 	 */
-	updateConversationTimestamp(): void {
-		if (!this.activeConversation) return;
+	updateConversationTimestamp(convId?: string): void {
+		const targetId = convId ?? this.activeConversation?.id;
+		if (!targetId) return;
 
-		const chatIndex = this.conversations.findIndex((c) => c.id === this.activeConversation!.id);
+		const now = Date.now();
+
+		const chatIndex = this.conversations.findIndex((c) => c.id === targetId);
 
 		if (chatIndex !== -1) {
-			this.conversations[chatIndex].lastModified = Date.now();
+			this.conversations[chatIndex].lastModified = now;
 			const updatedConv = this.conversations.splice(chatIndex, 1)[0];
 			this.conversations = [updatedConv, ...this.conversations];
 		}
+
+		if (this.activeConversation?.id === targetId) {
+			this.activeConversation = { ...this.activeConversation, lastModified: now };
+		}
+
+		DatabaseService.updateConversation(targetId, { lastModified: now }).catch((error) =>
+			console.error('Failed to update conversation timestamp:', error)
+		);
 	}
 
 	/**
@@ -773,7 +789,6 @@ class ConversationsStore {
 		if (convIndex !== -1) {
 			this.conversations[convIndex].mcpServerOverrides =
 				newOverrides.length > 0 ? newOverrides : undefined;
-			this.conversations = [...this.conversations];
 		}
 	}
 
@@ -837,7 +852,6 @@ class ConversationsStore {
 		const convIndex = this.conversations.findIndex((c) => c.id === this.activeConversation!.id);
 		if (convIndex !== -1) {
 			this.conversations[convIndex].reasoningEffort = effort;
-			this.conversations = [...this.conversations];
 		}
 	}
 
diff --git a/tools/ui/src/lib/utils/legacy-migration.ts b/tools/ui/src/lib/utils/legacy-migration.ts
deleted file mode 100644
index 6b0890a36..000000000
--- a/tools/ui/src/lib/utils/legacy-migration.ts
+++ /dev/null
@@ -1,364 +0,0 @@
-/**
- * @deprecated Legacy migration utility — remove at some point in the future once all users have migrated to the new structured agentic message format.
- *
- * Converts old marker-based agentic messages to the new structured format
- * with separate messages per turn.
- *
- * Old format: Single assistant message with markers in content:
- *   <<<reasoning_content_start>>>...<<<reasoning_content_end>>>
- *   <<<AGENTIC_TOOL_CALL_START>>>...<<<AGENTIC_TOOL_CALL_END>>>
- *
- * New format: Separate messages per turn:
- *   - assistant (content + reasoningContent + toolCalls)
- *   - tool (toolCallId + content)
- *   - assistant (next turn)
- *   - ...
- */
-
-import { LEGACY_AGENTIC_REGEX, LEGACY_REASONING_TAGS } from '$lib/constants';
-import { DatabaseService } from '$lib/services/database.service';
-import { MessageRole, MessageType } from '$lib/enums';
-import type { DatabaseMessage } from '$lib/types/database';
-
-const MIGRATION_DONE_KEY = 'llama-ui-migration-v2-done';
-/** @deprecated Use {@link MIGRATION_DONE_KEY} instead */
-const DEPRECATED_MIGRATION_DONE_KEY = 'llama-webui-migration-v2-done';
-
-/**
- * @deprecated Part of legacy migration — remove with the migration module.
- * Check if migration has been performed.
- */
-export function isMigrationNeeded(): boolean {
-	try {
-		// Check new key first, fall back to deprecated old key
-		if (localStorage.getItem(MIGRATION_DONE_KEY)) return false;
-		if (localStorage.getItem(DEPRECATED_MIGRATION_DONE_KEY)) {
-			// Migrate to new key
-			try {
-				localStorage.setItem(MIGRATION_DONE_KEY, String(Date.now()));
-				localStorage.removeItem(DEPRECATED_MIGRATION_DONE_KEY);
-			} catch {
-				// Ignore storage errors
-			}
-			return false;
-		}
-		return true;
-	} catch {
-		return false;
-	}
-}
-
-/**
- * Mark migration as done.
- */
-function markMigrationDone(): void {
-	try {
-		localStorage.setItem(MIGRATION_DONE_KEY, String(Date.now()));
-	} catch {
-		// Ignore localStorage errors
-	}
-}
-
-/**
- * Check if a message has legacy markers in its content.
- */
-function hasLegacyMarkers(message: DatabaseMessage): boolean {
-	if (!message.content) return false;
-	return LEGACY_AGENTIC_REGEX.HAS_LEGACY_MARKERS.test(message.content);
-}
-
-/**
- * Extract reasoning content from legacy marker format.
- */
-function extractLegacyReasoning(content: string): { reasoning: string; cleanContent: string } {
-	let reasoning = '';
-	let cleanContent = content;
-
-	// Extract all reasoning blocks
-	const re = new RegExp(LEGACY_AGENTIC_REGEX.REASONING_EXTRACT.source, 'g');
-	let match;
-	while ((match = re.exec(content)) !== null) {
-		reasoning += match[1];
-	}
-
-	// Remove reasoning tags from content
-	cleanContent = cleanContent
-		.replace(new RegExp(LEGACY_AGENTIC_REGEX.REASONING_BLOCK.source, 'g'), '')
-		.replace(LEGACY_AGENTIC_REGEX.REASONING_OPEN, '');
-
-	return { reasoning, cleanContent };
-}
-
-/**
- * Parse legacy content with tool call markers into structured turns.
- */
-interface ParsedTurn {
-	textBefore: string;
-	toolCalls: Array<{
-		name: string;
-		args: string;
-		result: string;
-	}>;
-}
-
-function parseLegacyToolCalls(content: string): ParsedTurn[] {
-	const turns: ParsedTurn[] = [];
-	const regex = new RegExp(LEGACY_AGENTIC_REGEX.COMPLETED_TOOL_CALL.source, 'g');
-
-	let lastIndex = 0;
-	let currentTurn: ParsedTurn = { textBefore: '', toolCalls: [] };
-	let match;
-
-	while ((match = regex.exec(content)) !== null) {
-		const textBefore = content.slice(lastIndex, match.index).trim();
-
-		// If there's text between tool calls and we already have tool calls,
-		// that means a new turn started (text after tool results = new LLM turn)
-		if (textBefore && currentTurn.toolCalls.length > 0) {
-			turns.push(currentTurn);
-			currentTurn = { textBefore, toolCalls: [] };
-		} else if (textBefore && currentTurn.toolCalls.length === 0) {
-			currentTurn.textBefore = textBefore;
-		}
-
-		currentTurn.toolCalls.push({
-			name: match[1],
-			args: match[2],
-			result: match[3].replace(/^\n+|\n+$/g, '')
-		});
-
-		lastIndex = match.index + match[0].length;
-	}
-
-	// Any remaining text after the last tool call
-	const remainingText = content.slice(lastIndex).trim();
-
-	if (currentTurn.toolCalls.length > 0) {
-		turns.push(currentTurn);
-	}
-
-	// If there's text after all tool calls, it's the final assistant response
-	if (remainingText) {
-		// Remove any partial/open markers
-		const cleanRemaining = remainingText
-			.replace(LEGACY_AGENTIC_REGEX.AGENTIC_TOOL_CALL_OPEN, '')
-			.trim();
-		if (cleanRemaining) {
-			turns.push({ textBefore: cleanRemaining, toolCalls: [] });
-		}
-	}
-
-	// If no tool calls found at all, return the original content as a single turn
-	if (turns.length === 0) {
-		turns.push({ textBefore: content.trim(), toolCalls: [] });
-	}
-
-	return turns;
-}
-
-/**
- * Migrate a single conversation's messages from legacy format to new format.
- */
-async function migrateConversation(convId: string): Promise<number> {
-	const allMessages = await DatabaseService.getConversationMessages(convId);
-	let migratedCount = 0;
-
-	for (const message of allMessages) {
-		if (message.role !== MessageRole.ASSISTANT) continue;
-		if (!hasLegacyMarkers(message)) {
-			// Still check for reasoning-only markers (no tool calls)
-			if (message.content?.includes(LEGACY_REASONING_TAGS.START)) {
-				const { reasoning, cleanContent } = extractLegacyReasoning(message.content);
-				await DatabaseService.updateMessage(message.id, {
-					content: cleanContent.trim(),
-					reasoningContent: reasoning || undefined
-				});
-				migratedCount++;
-			}
-			continue;
-		}
-
-		// Has agentic markers - full migration needed
-		const { reasoning, cleanContent } = extractLegacyReasoning(message.content);
-		const turns = parseLegacyToolCalls(cleanContent);
-
-		// Parse existing toolCalls JSON to try to match IDs
-		let existingToolCalls: Array<{
-			id: string;
-			function?: { name: string; arguments: string };
-		}> = [];
-		if (message.toolCalls) {
-			try {
-				existingToolCalls = JSON.parse(message.toolCalls);
-			} catch {
-				// Ignore
-			}
-		}
-
-		// First turn uses the existing message
-		const firstTurn = turns[0];
-		if (!firstTurn) continue;
-
-		// Match tool calls from the first turn to existing IDs
-		const firstTurnToolCalls = firstTurn.toolCalls.map((tc, i) => {
-			const existing =
-				existingToolCalls.find((e) => e.function?.name === tc.name) || existingToolCalls[i];
-			return {
-				id: existing?.id || `legacy_tool_${i}`,
-				type: 'function' as const,
-				function: { name: tc.name, arguments: tc.args }
-			};
-		});
-
-		// Update the existing message for the first turn
-		await DatabaseService.updateMessage(message.id, {
-			content: firstTurn.textBefore,
-			reasoningContent: reasoning || undefined,
-			toolCalls: firstTurnToolCalls.length > 0 ? JSON.stringify(firstTurnToolCalls) : ''
-		});
-
-		let currentParentId = message.id;
-		let toolCallIdCounter = existingToolCalls.length;
-
-		// Create tool result messages for the first turn
-		for (let i = 0; i < firstTurn.toolCalls.length; i++) {
-			const tc = firstTurn.toolCalls[i];
-			const toolCallId = firstTurnToolCalls[i]?.id || `legacy_tool_${i}`;
-
-			const toolMsg = await DatabaseService.createMessageBranch(
-				{
-					convId,
-					type: MessageType.TEXT,
-					role: MessageRole.TOOL,
-					content: tc.result,
-					toolCallId,
-					timestamp: message.timestamp + i + 1,
-					toolCalls: '',
-					children: []
-				},
-				currentParentId
-			);
-			currentParentId = toolMsg.id;
-		}
-
-		// Create messages for subsequent turns
-		for (let turnIdx = 1; turnIdx < turns.length; turnIdx++) {
-			const turn = turns[turnIdx];
-
-			const turnToolCalls = turn.toolCalls.map((tc, i) => {
-				const idx = toolCallIdCounter + i;
-				const existing = existingToolCalls[idx];
-				return {
-					id: existing?.id || `legacy_tool_${idx}`,
-					type: 'function' as const,
-					function: { name: tc.name, arguments: tc.args }
-				};
-			});
-			toolCallIdCounter += turn.toolCalls.length;
-
-			// Create assistant message for this turn
-			const assistantMsg = await DatabaseService.createMessageBranch(
-				{
-					convId,
-					type: MessageType.TEXT,
-					role: MessageRole.ASSISTANT,
-					content: turn.textBefore,
-					timestamp: message.timestamp + turnIdx * 100,
-					toolCalls: turnToolCalls.length > 0 ? JSON.stringify(turnToolCalls) : '',
-					children: [],
-					model: message.model
-				},
-				currentParentId
-			);
-			currentParentId = assistantMsg.id;
-
-			// Create tool result messages for this turn
-			for (let i = 0; i < turn.toolCalls.length; i++) {
-				const tc = turn.toolCalls[i];
-				const toolCallId = turnToolCalls[i]?.id || `legacy_tool_${toolCallIdCounter + i}`;
-
-				const toolMsg = await DatabaseService.createMessageBranch(
-					{
-						convId,
-						type: MessageType.TEXT,
-						role: MessageRole.TOOL,
-						content: tc.result,
-						toolCallId,
-						timestamp: message.timestamp + turnIdx * 100 + i + 1,
-						toolCalls: '',
-						children: []
-					},
-					currentParentId
-				);
-				currentParentId = toolMsg.id;
-			}
-		}
-
-		// Re-parent any children of the original message to the last created message
-		// (the original message's children list was the next user message or similar)
-		if (message.children.length > 0 && currentParentId !== message.id) {
-			for (const childId of message.children) {
-				// Skip children we just created (they were already properly parented)
-				const child = allMessages.find((m) => m.id === childId);
-				if (!child) continue;
-				// Only re-parent non-tool messages that were original children
-				if (child.role !== MessageRole.TOOL) {
-					await DatabaseService.updateMessage(childId, { parent: currentParentId });
-					// Add to new parent's children
-					const newParent = await DatabaseService.getConversationMessages(convId).then((msgs) =>
-						msgs.find((m) => m.id === currentParentId)
-					);
-					if (newParent && !newParent.children.includes(childId)) {
-						await DatabaseService.updateMessage(currentParentId, {
-							children: [...newParent.children, childId]
-						});
-					}
-				}
-			}
-			// Clear re-parented children from the original message
-			await DatabaseService.updateMessage(message.id, { children: [] });
-		}
-
-		migratedCount++;
-	}
-
-	return migratedCount;
-}
-
-/**
- * @deprecated Part of legacy migration — remove with the migration module.
- * Run the full migration across all conversations.
- * This should be called once at app startup if migration is needed.
- */
-export async function runLegacyMigration(): Promise<void> {
-	if (!isMigrationNeeded()) return;
-
-	if (import.meta.env.DEV && import.meta.env.VITE_DEBUG)
-		console.log('[Migration] Starting legacy message format migration...');
-
-	try {
-		const conversations = await DatabaseService.getAllConversations();
-		let totalMigrated = 0;
-
-		for (const conv of conversations) {
-			const count = await migrateConversation(conv.id);
-			totalMigrated += count;
-		}
-
-		if (import.meta.env.DEV && import.meta.env.VITE_DEBUG) {
-			if (totalMigrated > 0) {
-				console.log(
-					`[Migration] Migrated ${totalMigrated} messages across ${conversations.length} conversations`
-				);
-			} else {
-				console.log('[Migration] No legacy messages found, marking as done');
-			}
-		}
-
-		markMigrationDone();
-	} catch (error) {
-		console.error('[Migration] Failed to migrate legacy messages:', error);
-		// Still mark as done to avoid infinite retry loops
-		markMigrationDone();
-	}
-}

From d0bfb1981266c271cd0536a8aa7c5e863e7cdf61 Mon Sep 17 00:00:00 2001
From: Niklas Wenzel <dev@nikwen.de>
Date: Thu, 30 Jul 2026 10:11:27 +0200
Subject: [PATCH 08/30] metal: fix memory unwire if model is freed without any
 GPU operations (#26082)

* metal: fix memory leak if model is freed without any GPU operations

* metal: run dummy work only if residency sets are used

* metal: wrap function in #if defined

* metal: measure system-wide wired memory in test

* metal: always build regression test

Co-authored-by: YiChen Lv <63285796+forforever73@users.noreply.github.com>

---------

Co-authored-by: YiChen Lv <63285796+forforever73@users.noreply.github.com>
---
 ggml/src/ggml-metal/ggml-metal-device.h |  2 +-
 ggml/src/ggml-metal/ggml-metal-device.m | 39 +++++++++++++++++-
 tests/CMakeLists.txt                    |  3 ++
 tests/test-rset-release.cpp             | 53 +++++++++++++++++++++++++
 4 files changed, 94 insertions(+), 3 deletions(-)
 create mode 100644 tests/test-rset-release.cpp

diff --git a/ggml/src/ggml-metal/ggml-metal-device.h b/ggml/src/ggml-metal/ggml-metal-device.h
index d0956df50..91b841b67 100644
--- a/ggml/src/ggml-metal/ggml-metal-device.h
+++ b/ggml/src/ggml-metal/ggml-metal-device.h
@@ -213,7 +213,7 @@ typedef void * ggml_metal_rset_t;
 // a collection of residency sets (non-owning)
 typedef struct ggml_metal_rsets * ggml_metal_rsets_t;
 
-ggml_metal_rsets_t ggml_metal_rsets_init(void);
+ggml_metal_rsets_t ggml_metal_rsets_init(ggml_metal_device_t dev);
 void ggml_metal_rsets_free(ggml_metal_rsets_t rsets);
 
 //
diff --git a/ggml/src/ggml-metal/ggml-metal-device.m b/ggml/src/ggml-metal/ggml-metal-device.m
index 4edd77c6f..7d2a68685 100644
--- a/ggml/src/ggml-metal/ggml-metal-device.m
+++ b/ggml/src/ggml-metal/ggml-metal-device.m
@@ -557,7 +557,32 @@ struct ggml_metal_rsets {
     dispatch_group_t d_group;
 };
 
-ggml_metal_rsets_t ggml_metal_rsets_init(void) {
+#if defined(GGML_METAL_HAS_RESIDENCY_SETS)
+static void ggml_metal_dummy_work(ggml_metal_device_t dev) {
+    if (dev->mtl_queue == nil) {
+        return;
+    }
+
+    @autoreleasepool {
+        // perform a minimal dummy operation on the GPU
+        id<MTLBuffer> buf = [dev->mtl_device newBufferWithLength:1 options:MTLResourceStorageModePrivate];
+        id<MTLCommandBuffer> cmd_buf = [dev->mtl_queue commandBuffer];
+
+        {
+            id<MTLBlitCommandEncoder> encoder = [cmd_buf blitCommandEncoder];
+
+            [encoder fillBuffer:buf range:NSMakeRange(0, 1) value:0];
+
+            [encoder endEncoding];
+        }
+
+        [cmd_buf commit];
+        [buf release];
+    }
+}
+#endif
+
+ggml_metal_rsets_t ggml_metal_rsets_init(ggml_metal_device_t dev) {
     ggml_metal_rsets_t res = calloc(1, sizeof(struct ggml_metal_rsets));
 
     res->lock = [[NSLock alloc] init];
@@ -610,6 +635,15 @@ ggml_metal_rsets_t ggml_metal_rsets_init(void) {
 #endif
     });
 
+#if defined(GGML_METAL_HAS_RESIDENCY_SETS)
+    if (@available(macOS 15.0, iOS 18.0, tvOS 18.0, visionOS 2.0, *)) {
+        // workaround for residency set memory not being released if no GPU operation occurs
+        // https://developer.apple.com/forums/thread/839089
+        // https://github.com/ggml-org/llama.cpp/issues/25937
+        ggml_metal_dummy_work(dev);
+    }
+#endif
+
     return res;
 }
 
@@ -864,7 +898,7 @@ ggml_metal_device_t ggml_metal_device_init(int device) {
             }
 
             if (dev->props.use_residency_sets) {
-                dev->rsets = ggml_metal_rsets_init();
+                dev->rsets = ggml_metal_rsets_init(dev);
             } else {
                 dev->rsets = nil;
             }
@@ -1484,6 +1518,7 @@ static void ggml_metal_buffer_rset_free(ggml_metal_buffer_t buf) {
         if (buf->rset) {
             [buf->rset endResidency];
             [buf->rset removeAllAllocations];
+            [buf->rset commit];
             [buf->rset release];
         }
     }
diff --git a/tests/CMakeLists.txt b/tests/CMakeLists.txt
index 7a93b19a0..805b74472 100644
--- a/tests/CMakeLists.txt
+++ b/tests/CMakeLists.txt
@@ -278,6 +278,9 @@ set_tests_properties(test-state-restore-fragmented PROPERTIES FIXTURES_REQUIRED
 llama_build_and_test(test-save-load-state.cpp LABEL "model" ARGS -m "${MODEL_DEST}")
 set_tests_properties(test-save-load-state PROPERTIES FIXTURES_REQUIRED test-download-model)
 
+if (APPLE)
+    llama_build(test-rset-release.cpp get-model.cpp)
+endif()
 if (NOT GGML_BACKEND_DL)
     # these tests use the backends directly and cannot be built with dynamic loading
     llama_build_and_test(test-barrier.cpp)
diff --git a/tests/test-rset-release.cpp b/tests/test-rset-release.cpp
new file mode 100644
index 000000000..bf03c5e8b
--- /dev/null
+++ b/tests/test-rset-release.cpp
@@ -0,0 +1,53 @@
+// ref: https://github.com/ggml-org/llama.cpp/issues/25937
+// only works reliably when run with a large model that occupies 3GB+ of wired memory
+// thus, this test is not run by default
+// example model to run with: google/gemma-4-E4B-it-qat-q4_0-gguf
+
+#include <cstdint>
+#include <mach/mach.h>
+#include <mach/mach_host.h>
+#include <unistd.h>
+
+#include "llama.h"
+#include "get-model.h"
+
+static uint64_t wired_memory() {
+    vm_statistics64_data_t vmstat;
+    mach_msg_type_number_t count = HOST_VM_INFO64_COUNT;
+    if (host_statistics64(mach_host_self(), HOST_VM_INFO64, (host_info64_t)&vmstat, &count) != KERN_SUCCESS) {
+        return UINT64_MAX;
+    }
+    return static_cast<uint64_t>(vmstat.wire_count) * vm_kernel_page_size;
+}
+
+int main(int argc, char ** argv) {
+    auto * model_path = get_model_or_exit(argc, argv);
+
+    llama_backend_init();
+
+    const uint64_t wired_initial = wired_memory();
+
+    llama_model_params params = llama_model_default_params();
+    params.load_mode = LLAMA_LOAD_MODE_NONE;
+    struct llama_model* model = llama_model_load_from_file(model_path, params);
+
+    const uint64_t wired_loaded = wired_memory();
+    const uint64_t wired_delta = wired_loaded - wired_initial;
+    // system memory fluctuates, so we need to allocate enough to reliably detect the release
+    GGML_ASSERT(wired_delta > 2'000'000'000); // 2GB
+
+    llama_model_free(model);
+
+    const uint64_t t_start_ms = ggml_time_ms();
+
+    // expect most of the allocated memory to be released within 10 seconds
+    // we allow for some tolerance due to system-wide memory fluctuations
+    while (wired_memory() > wired_loaded - 0.75 * wired_delta) {
+        GGML_ASSERT(ggml_time_ms() - t_start_ms < 10'000);
+        usleep(100'000); // 100ms
+    }
+
+    llama_backend_free();
+
+    return 0;
+}

From b2f221684fcd898e947a121baeda80f345da3e6b Mon Sep 17 00:00:00 2001
From: timkhronos <timkhronos@gmail.com>
Date: Thu, 30 Jul 2026 10:30:18 +0200
Subject: [PATCH 09/30] Remove custom cpu op from the M3 graph, express with
 stock ops (#26297)

---
 src/models/minimax-m3.cpp | 93 +++++++++------------------------------
 1 file changed, 21 insertions(+), 72 deletions(-)

diff --git a/src/models/minimax-m3.cpp b/src/models/minimax-m3.cpp
index 6068fc6b8..3e7bada64 100644
--- a/src/models/minimax-m3.cpp
+++ b/src/models/minimax-m3.cpp
@@ -2,7 +2,6 @@
 #include "llama-kv-cache.h"
 #include <cmath>
 #include <vector>
-#include <algorithm>
 #include <cstdint>
 
 // MiniMax-M3: MiniMax-M2 style GQA (per-head QK-norm, partial rotary) with
@@ -126,68 +125,6 @@ public:
     int64_t nblk;
 };
 
-// pooled score of a block with no visible token: -inf from the mask, or -FLT_MAX from the
-// max-pool identity when every element of the block is -inf
-static inline bool msa_score_masked(float x) { return x <= -1e30f; }
-
-// MSA block selection (batch regime)
-// CPU custom op, the token-level expansion and the combination with the causal mask happen on the GPU.
-static void msa_block_mask_op(struct ggml_tensor * dst, int ith, int nth, void * userdata) {
-    const struct ggml_tensor * bs   = dst->src[0];
-    const struct ggml_tensor * bias = dst->src[1];
-    const msa_params * p = (const msa_params *) userdata;
-
-    const int nblk = (int) bs->ne[0];
-    const int Hd   = (int) bs->ne[1];
-    const int S    = (int) bs->ne[2];
-
-    GGML_ASSERT(bs->type   == GGML_TYPE_F32 && ggml_is_contiguous(bs));
-    GGML_ASSERT(bias->type == GGML_TYPE_F32 && ggml_is_contiguous(bias));
-    GGML_ASSERT(dst->type  == GGML_TYPE_F16 && ggml_is_contiguous(dst));
-    GGML_ASSERT(dst->ne[0] == nblk && dst->ne[1] == S && dst->ne[2] == Hd);
-    GGML_ASSERT(bias->ne[0] == nblk && bias->ne[1] == S);
-
-    const int topk = p->topk_blocks < nblk ? p->topk_blocks : nblk;
-
-    const ggml_fp16_t f16_zero = ggml_fp32_to_fp16(0.0f);
-    const ggml_fp16_t f16_ninf = ggml_fp32_to_fp16(-INFINITY);
-
-    std::vector<float> rank(nblk);
-    std::vector<char>  valid(nblk);
-    std::vector<int>   ord(nblk);
-
-    ggml_fp16_t * out = (ggml_fp16_t *) dst->data;
-
-    for (int i = ith; i < S; i += nth) {
-        const float * bias_col = (const float *) bias->data + (size_t) i * nblk;
-        for (int h = 0; h < Hd; ++h) {
-            const float * bs_col = (const float *) bs->data + ((size_t) i * Hd + h) * nblk;
-
-            for (int bk = 0; bk < nblk; ++bk) {
-                // a block is selectable if it has a visible token or is locally forced
-                valid[bk] = !msa_score_masked(bs_col[bk]) || bias_col[bk] > 0.0f;
-                rank [bk] = bias_col[bk] > 0.0f ? bias_col[bk] : bs_col[bk];
-                ord  [bk] = bk;
-            }
-
-            std::partial_sort(ord.begin(), ord.begin() + topk, ord.end(),
-                              [&](int a, int b) { return rank[a] > rank[b]; });
-
-            ggml_fp16_t * dst_col = out + ((size_t) h * S + i) * nblk;
-            for (int bk = 0; bk < nblk; ++bk) {
-                dst_col[bk] = f16_ninf;
-            }
-            for (int t = 0; t < topk; ++t) {
-                const int bk = ord[t];
-                if (!valid[bk]) {
-                    break;   // sorted desc: first invalid -> fewer than topk selectable blocks
-                }
-                dst_col[bk] = f16_zero;
-            }
-        }
-    }
-}
-
 // One FA call for all GQA groups (and at multi-stream decode, all streams) by mapping them onto the FA sequence dim (ne[3])
 ggml_tensor * llama_model_minimax_m3::graph::build_attn_msa_fa(
         ggml_tensor * q_cur,   // [D, HQ, T]
@@ -433,8 +370,8 @@ llama_model_minimax_m3::graph::graph(const llama_model & model, const llm_graph_
                                 msa_mf->nb[1], msa_mf->nb[1], st*msa_mf->nb[3]);
                         ggml_tensor * km_s = ggml_view_3d(ctx0, msa_kqm, n_kv, n_tps, 1,
                                 msa_kqm->nb[1], msa_kqm->nb[3], st*msa_kqm->nb[3]);
-                        ggml_tensor * bias_s = ggml_view_2d(ctx0, msa_loc->bias, nblk, n_tps,
-                                msa_loc->bias->nb[1], st*n_tps*msa_loc->bias->nb[1]);
+                        ggml_tensor * bias_s = ggml_view_3d(ctx0, msa_loc->bias, nblk, 1, n_tps,
+                                msa_loc->bias->nb[1], msa_loc->bias->nb[1], st*n_tps*msa_loc->bias->nb[1]);
                         ggml_tensor * q_s = ggml_view_3d(ctx0, Qcur, D, n_head, n_tps,
                                 Qcur->nb[1], Qcur->nb[2], st*n_tps*Qcur->nb[2]);
                         ggml_tensor * k_s = ggml_view_4d(ctx0, k, D, HKV, n_kv, 1,
@@ -453,15 +390,27 @@ llama_model_minimax_m3::graph::graph(const llama_model & model, const llm_graph_
                         ggml_tensor * bs = ggml_pool_2d(ctx0, sc, GGML_OP_POOL_MAX, blk, 1, blk, 1, 0, 0);
                         cb(bs, "msa_bs", il);
 
-                        // block-level 0/-inf keep mask on the CPU, tiny transfer
-                        ggml_tensor * srcs[2] = { bs, bias_s };
-                        ggml_tensor * bm = ggml_custom_4d(ctx0, GGML_TYPE_F16,
-                                nblk, n_tps, Hd, 1,
-                                srcs, 2, msa_block_mask_op, GGML_N_TASKS_MAX,
-                                const_cast<msa_params *>(&mm.msa_p));
+                        // bias the scores so locally-forced blocks always rank first
+                        ggml_tensor * bsf = ggml_add(ctx0, bs, bias_s);   // [nblk, Hd, n_tps]
+                        cb(bsf, "msa_bsf", il);
+
+                        ggml_tensor * idx = ggml_top_k(ctx0, bsf, K);   // [K, Hd, n_tps] i32
+
+                        ggml_tensor * ninf = ggml_cast(ctx0,
+                                ggml_scale_bias(ctx0, bias_s, 0.0f, -1e30f),
+                                GGML_TYPE_F16);                              // [nblk, 1, n_tps]
+                        ninf = ggml_repeat_4d(ctx0, ninf, nblk, Hd, n_tps, 1);
+                        ggml_tensor * zero = ggml_scale(ctx0,
+                                ggml_cast(ctx0, idx, GGML_TYPE_F32), 0.0f);
+                        ggml_tensor * bm = ggml_set_rows(ctx0,
+                                ggml_reshape_3d(ctx0, ninf, 1, nblk, Hd*n_tps),
+                                ggml_reshape_3d(ctx0, zero, 1, K,    Hd*n_tps),
+                                ggml_reshape_2d(ctx0, idx,     K,    Hd*n_tps));
+                        bm = ggml_reshape_3d(ctx0, bm, nblk, Hd, n_tps);
+                        bm = ggml_cont(ctx0, ggml_permute(ctx0, bm, 0, 2, 1, 3)); // [nblk, n_tps, Hd]
                         cb(bm, "msa_block_mask", il);
 
-                        // expand block -> token granularity on the GPU (j = bk*blk + t),
+                        // expand block -> token granularity (j = bk*blk + t),
                         // then combine with the causal mask in place
                         ggml_tensor * bmx = ggml_repeat_4d(ctx0,
                                 ggml_reshape_3d(ctx0, bm, 1, nblk, n_tps*Hd),

From 9b2a088819cda774bdbf713168ee1eee8498cda5 Mon Sep 17 00:00:00 2001
From: Pasha Khosravi <khosravipasha@users.noreply.github.com>
Date: Thu, 30 Jul 2026 02:33:25 -0700
Subject: [PATCH 10/30] CUDA: add Q2_0 support (#25707)

---
 ggml/src/ggml-cuda/common.cuh                 |  7 ++
 ggml/src/ggml-cuda/convert.cu                 | 12 +++
 ggml/src/ggml-cuda/dequantize.cuh             | 20 ++++
 ggml/src/ggml-cuda/getrows.cu                 |  4 +
 ggml/src/ggml-cuda/ggml-cuda.cu               |  2 +
 ggml/src/ggml-cuda/mmq-config-ampere.cuh      | 17 ++++
 ggml/src/ggml-cuda/mmq-config-cdna.cuh        |  8 ++
 ggml/src/ggml-cuda/mmq-config-pascal.cuh      | 12 +++
 ggml/src/ggml-cuda/mmq-config-rdna2.cuh       | 12 +++
 ggml/src/ggml-cuda/mmq-config-rdna3-5.cuh     | 12 +++
 ggml/src/ggml-cuda/mmq-config-rdna3.cuh       | 12 +++
 ggml/src/ggml-cuda/mmq-config-rdna4.cuh       | 12 +++
 ggml/src/ggml-cuda/mmq-load-tiles.cuh         | 97 +++++++++++++++++++
 ggml/src/ggml-cuda/mmq.cu                     |  4 +
 ggml/src/ggml-cuda/mmq.cuh                    | 15 +++
 ggml/src/ggml-cuda/mmvq.cu                    |  8 ++
 .../template-instances/generate_cu_files.py   |  1 +
 .../template-instances/mmq-instance-q2_0.cu   |  5 +
 ggml/src/ggml-cuda/vecdotq.cuh                | 61 ++++++++++++
 tests/test-backend-ops.cpp                    |  5 +
 20 files changed, 326 insertions(+)
 create mode 100644 ggml/src/ggml-cuda/template-instances/mmq-instance-q2_0.cu

diff --git a/ggml/src/ggml-cuda/common.cuh b/ggml/src/ggml-cuda/common.cuh
index fa2dd2620..33be16dc5 100644
--- a/ggml/src/ggml-cuda/common.cuh
+++ b/ggml/src/ggml-cuda/common.cuh
@@ -977,6 +977,13 @@ struct ggml_cuda_type_traits<GGML_TYPE_Q1_0> {
     static constexpr int qi = QI1_0;
 };
 
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_Q2_0> {
+    static constexpr int qk = QK2_0;
+    static constexpr int qr = QR2_0;
+    static constexpr int qi = QI2_0;
+};
+
 template<>
 struct ggml_cuda_type_traits<GGML_TYPE_Q4_0> {
     static constexpr int qk = QK4_0;
diff --git a/ggml/src/ggml-cuda/convert.cu b/ggml/src/ggml-cuda/convert.cu
index 946e02af5..360c614a4 100644
--- a/ggml/src/ggml-cuda/convert.cu
+++ b/ggml/src/ggml-cuda/convert.cu
@@ -459,6 +459,8 @@ to_bf16_cuda_t ggml_get_to_bf16_cuda(ggml_type type) {
     switch (type) {
         case GGML_TYPE_Q1_0:
             return dequantize_block_cont_cuda<QK1_0, QR1_0, dequantize_q1_0>;
+        case GGML_TYPE_Q2_0:
+            return dequantize_block_cont_cuda<QK2_0, QR2_0, dequantize_q2_0>;
         case GGML_TYPE_Q4_0:
             return dequantize_row_q4_0_cuda;
         case GGML_TYPE_Q4_1:
@@ -514,6 +516,8 @@ to_fp16_cuda_t ggml_get_to_fp16_cuda(ggml_type type) {
     switch (type) {
         case GGML_TYPE_Q1_0:
             return dequantize_block_cont_cuda<QK1_0, QR1_0, dequantize_q1_0>;
+        case GGML_TYPE_Q2_0:
+            return dequantize_block_cont_cuda<QK2_0, QR2_0, dequantize_q2_0>;
         case GGML_TYPE_Q4_0:
             return dequantize_row_q4_0_cuda;
         case GGML_TYPE_Q4_1:
@@ -572,6 +576,8 @@ to_fp32_cuda_t ggml_get_to_fp32_cuda(ggml_type type) {
     switch (type) {
         case GGML_TYPE_Q1_0:
             return dequantize_block_cont_cuda<QK1_0, QR1_0, dequantize_q1_0>;
+        case GGML_TYPE_Q2_0:
+            return dequantize_block_cont_cuda<QK2_0, QR2_0, dequantize_q2_0>;
         case GGML_TYPE_Q4_0:
             return dequantize_row_q4_0_cuda;
         case GGML_TYPE_Q4_1:
@@ -629,6 +635,8 @@ to_fp16_nc_cuda_t ggml_get_to_fp16_nc_cuda(ggml_type type) {
             return convert_unary_cuda<float>;
         case GGML_TYPE_Q1_0:
             return dequantize_block_cuda<QK1_0, QR1_0, dequantize_q1_0>;
+        case GGML_TYPE_Q2_0:
+            return dequantize_block_cuda<QK2_0, QR2_0, dequantize_q2_0>;
         case GGML_TYPE_Q4_0:
             return dequantize_block_cuda<QK4_0, QR4_0, dequantize_q4_0>;
         case GGML_TYPE_Q4_1:
@@ -652,6 +660,8 @@ to_bf16_nc_cuda_t ggml_get_to_bf16_nc_cuda(ggml_type type) {
             return convert_unary_cuda<float, nv_bfloat16>;
         case GGML_TYPE_Q1_0:
             return dequantize_block_cuda<QK1_0, QR1_0, dequantize_q1_0>;
+        case GGML_TYPE_Q2_0:
+            return dequantize_block_cuda<QK2_0, QR2_0, dequantize_q2_0>;
         case GGML_TYPE_Q4_0:
             return dequantize_block_cuda<QK4_0, QR4_0, dequantize_q4_0>;
         case GGML_TYPE_Q4_1:
@@ -675,6 +685,8 @@ to_fp32_nc_cuda_t ggml_get_to_fp32_nc_cuda(ggml_type type) {
             return convert_unary_cuda<half, float>;
         case GGML_TYPE_Q1_0:
             return dequantize_block_cuda<QK1_0, QR1_0, dequantize_q1_0>;
+        case GGML_TYPE_Q2_0:
+            return dequantize_block_cuda<QK2_0, QR2_0, dequantize_q2_0>;
         case GGML_TYPE_Q4_0:
             return dequantize_block_cuda<QK4_0, QR4_0, dequantize_q4_0>;
         case GGML_TYPE_Q4_1:
diff --git a/ggml/src/ggml-cuda/dequantize.cuh b/ggml/src/ggml-cuda/dequantize.cuh
index 8ab5ad8e7..651524be7 100644
--- a/ggml/src/ggml-cuda/dequantize.cuh
+++ b/ggml/src/ggml-cuda/dequantize.cuh
@@ -23,6 +23,26 @@ static __device__ __forceinline__ void dequantize_q1_0(const void * vx, const in
     v.y = (2*bit_1 - 1) * d;
 }
 
+static __device__ __forceinline__ void dequantize_q2_0(const void * vx, const int64_t ib, const int iqs, float2 & v){
+    const block_q2_0 * x = (const block_q2_0 *) vx;
+
+    const float d = x[ib].d;
+
+    // Q2_0: 2 bits per element, 4 elements per byte.
+    // Stored code c in {0,1,2,3} maps to symbol s = c - 1 in {-1, 0, +1, +2}.
+    const int byte_index_0 = iqs / 4;
+    const int bit_offset_0 = (iqs % 4) * 2;
+
+    const int byte_index_1 = (iqs + 1) / 4;
+    const int bit_offset_1 = ((iqs + 1) % 4) * 2;
+
+    const int c0 = (x[ib].qs[byte_index_0] >> bit_offset_0) & 0x3;
+    const int c1 = (x[ib].qs[byte_index_1] >> bit_offset_1) & 0x3;
+
+    v.x = (c0 - 1) * d;
+    v.y = (c1 - 1) * d;
+}
+
 static __device__ __forceinline__ void dequantize_q4_0(const void * vx, const int64_t ib, const int iqs, float2 & v){
     const block_q4_0 * x = (const block_q4_0 *) vx;
 
diff --git a/ggml/src/ggml-cuda/getrows.cu b/ggml/src/ggml-cuda/getrows.cu
index a9ec4f697..6b36b8fbc 100644
--- a/ggml/src/ggml-cuda/getrows.cu
+++ b/ggml/src/ggml-cuda/getrows.cu
@@ -320,6 +320,10 @@ static void ggml_cuda_get_rows_switch_src0_type(
             get_rows_cuda_q<QK1_0, QR1_0, dequantize_q1_0>(src0_d, src1_d, dst_d,
                 ne00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb1, nb2, nb3, stream);
             break;
+        case GGML_TYPE_Q2_0:
+            get_rows_cuda_q<QK2_0, QR2_0, dequantize_q2_0>(src0_d, src1_d, dst_d,
+                ne00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb1, nb2, nb3, stream);
+            break;
         case GGML_TYPE_Q4_0:
             get_rows_cuda_q<QK4_0, QR4_0, dequantize_q4_0>(src0_d, src1_d, dst_d,
                 ne00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb1, nb2, nb3, stream);
diff --git a/ggml/src/ggml-cuda/ggml-cuda.cu b/ggml/src/ggml-cuda/ggml-cuda.cu
index e73a7b890..d22b5657d 100644
--- a/ggml/src/ggml-cuda/ggml-cuda.cu
+++ b/ggml/src/ggml-cuda/ggml-cuda.cu
@@ -4802,6 +4802,7 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
                     case GGML_TYPE_F32:
                     case GGML_TYPE_F16:
                     case GGML_TYPE_Q1_0:
+                    case GGML_TYPE_Q2_0:
                     case GGML_TYPE_Q4_0:
                     case GGML_TYPE_Q4_1:
                     case GGML_TYPE_Q5_0:
@@ -4840,6 +4841,7 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
                     case GGML_TYPE_BF16:
                     case GGML_TYPE_I32:
                     case GGML_TYPE_Q1_0:
+                    case GGML_TYPE_Q2_0:
                     case GGML_TYPE_Q4_0:
                     case GGML_TYPE_Q4_1:
                     case GGML_TYPE_Q5_0:
diff --git a/ggml/src/ggml-cuda/mmq-config-ampere.cuh b/ggml/src/ggml-cuda/mmq-config-ampere.cuh
index 0037bac3d..9f9fd1973 100644
--- a/ggml/src/ggml-cuda/mmq-config-ampere.cuh
+++ b/ggml/src/ggml-cuda/mmq-config-ampere.cuh
@@ -16,6 +16,23 @@ static constexpr __host__ __device__ ggml_cuda_mmq_config ggml_cuda_mmq_get_conf
     CASE(GGML_TYPE_Q1_0, 256, 1, 128, 112, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
     CASE(GGML_TYPE_Q1_0, 256, 1, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
 
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,   8, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,   8, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,  24, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,  40, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,  48, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,  80, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128,  96, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128, 112, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 256, 1, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+
     CASE(GGML_TYPE_Q4_0, 256, 1, 128,   8, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
     CASE(GGML_TYPE_Q4_0, 256, 1, 128,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
     CASE(GGML_TYPE_Q4_0, 256, 1, 128,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
diff --git a/ggml/src/ggml-cuda/mmq-config-cdna.cuh b/ggml/src/ggml-cuda/mmq-config-cdna.cuh
index 46ec6aa9d..4a8d89f72 100644
--- a/ggml/src/ggml-cuda/mmq-config-cdna.cuh
+++ b/ggml/src/ggml-cuda/mmq-config-cdna.cuh
@@ -7,6 +7,14 @@ static constexpr __host__ __device__ ggml_cuda_mmq_config ggml_cuda_mmq_get_conf
     CASE(GGML_TYPE_Q1_0, 512, 1, 128,  48, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
     CASE(GGML_TYPE_Q1_0, 512, 1, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
 
+    CASE(GGML_TYPE_Q2_0, 512, 1, 128,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
+    CASE(GGML_TYPE_Q2_0, 512, 1, 128,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
+    CASE(GGML_TYPE_Q2_0, 512, 1, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
+    CASE(GGML_TYPE_Q2_0, 512, 1, 128,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 512, 1, 128,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 512, 1, 128,  48, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+    CASE(GGML_TYPE_Q2_0, 512, 1, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, false);
+
     CASE(GGML_TYPE_Q4_0, 512, 1, 128,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
     CASE(GGML_TYPE_Q4_0, 512, 1, 128,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
     CASE(GGML_TYPE_Q4_0, 512, 1, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, true, true);
diff --git a/ggml/src/ggml-cuda/mmq-config-pascal.cuh b/ggml/src/ggml-cuda/mmq-config-pascal.cuh
index 8f0faac88..e7d4a9a3f 100644
--- a/ggml/src/ggml-cuda/mmq-config-pascal.cuh
+++ b/ggml/src/ggml-cuda/mmq-config-pascal.cuh
@@ -11,6 +11,18 @@ static constexpr __host__ __device__ ggml_cuda_mmq_config ggml_cuda_mmq_get_conf
     CASE(GGML_TYPE_Q1_0, 256, 2, 64,  48, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
     CASE(GGML_TYPE_Q1_0, 256, 2, 64,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
 
+    CASE(GGML_TYPE_Q2_0, 256, 2, 64,   8, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 64,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 64,   8, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 64,  24, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 64,  40, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 64,  48, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 64,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+
     CASE(GGML_TYPE_Q4_0, 256, 2, 64,   8, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
     CASE(GGML_TYPE_Q4_0, 256, 2, 64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
     CASE(GGML_TYPE_Q4_0, 256, 2, 64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
diff --git a/ggml/src/ggml-cuda/mmq-config-rdna2.cuh b/ggml/src/ggml-cuda/mmq-config-rdna2.cuh
index de4db0a3d..8324d9e1a 100644
--- a/ggml/src/ggml-cuda/mmq-config-rdna2.cuh
+++ b/ggml/src/ggml-cuda/mmq-config-rdna2.cuh
@@ -11,6 +11,18 @@ static constexpr __host__ __device__ ggml_cuda_mmq_config ggml_cuda_mmq_get_conf
     CASE(GGML_TYPE_Q1_0, 256, 2, 128,  48, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
     CASE(GGML_TYPE_Q1_0, 256, 2, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
 
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,   8, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,   8, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  24, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  40, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  48, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+
     CASE(GGML_TYPE_Q4_0, 256, 2, 128,   8, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
     CASE(GGML_TYPE_Q4_0, 256, 2, 128,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
     CASE(GGML_TYPE_Q4_0, 256, 2, 128,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
diff --git a/ggml/src/ggml-cuda/mmq-config-rdna3-5.cuh b/ggml/src/ggml-cuda/mmq-config-rdna3-5.cuh
index e420a32f0..180b2d937 100644
--- a/ggml/src/ggml-cuda/mmq-config-rdna3-5.cuh
+++ b/ggml/src/ggml-cuda/mmq-config-rdna3-5.cuh
@@ -11,6 +11,18 @@ static constexpr __host__ __device__ ggml_cuda_mmq_config ggml_cuda_mmq_get_conf
     CASE(GGML_TYPE_Q1_0, 256, 2, 128,  96, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
     CASE(GGML_TYPE_Q1_0, 256, 2, 128, 112, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
     CASE(GGML_TYPE_Q1_0, 256, 2, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  48, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  80, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  96, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128, 112, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
 
     CASE(GGML_TYPE_Q4_0, 128, 2,  64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
     CASE(GGML_TYPE_Q4_0, 128, 2,  64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
diff --git a/ggml/src/ggml-cuda/mmq-config-rdna3.cuh b/ggml/src/ggml-cuda/mmq-config-rdna3.cuh
index 122623060..676f27fea 100644
--- a/ggml/src/ggml-cuda/mmq-config-rdna3.cuh
+++ b/ggml/src/ggml-cuda/mmq-config-rdna3.cuh
@@ -11,6 +11,18 @@ static constexpr __host__ __device__ ggml_cuda_mmq_config ggml_cuda_mmq_get_conf
     CASE(GGML_TYPE_Q1_0, 256, 2, 128,  96, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
     CASE(GGML_TYPE_Q1_0, 256, 2, 128, 112, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
     CASE(GGML_TYPE_Q1_0, 256, 2, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  48, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  80, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  96, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128, 112, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
 
     CASE(GGML_TYPE_Q4_0, 128, 2,  64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
     CASE(GGML_TYPE_Q4_0, 128, 2,  64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
diff --git a/ggml/src/ggml-cuda/mmq-config-rdna4.cuh b/ggml/src/ggml-cuda/mmq-config-rdna4.cuh
index a224ecafc..9293d9d55 100644
--- a/ggml/src/ggml-cuda/mmq-config-rdna4.cuh
+++ b/ggml/src/ggml-cuda/mmq-config-rdna4.cuh
@@ -11,6 +11,18 @@ static constexpr __host__ __device__ ggml_cuda_mmq_config ggml_cuda_mmq_get_conf
     CASE(GGML_TYPE_Q1_0, 256, 2, 128,  96, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
     CASE(GGML_TYPE_Q1_0, 256, 2, 128, 112, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
     CASE(GGML_TYPE_Q1_0, 256, 2, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  48, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 128, 2,  64,  64, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  80, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128,  96, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128, 112, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
+    CASE(GGML_TYPE_Q2_0, 256, 2, 128, 128, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, false);
 
     CASE(GGML_TYPE_Q4_0, 128, 2,  64,  16, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
     CASE(GGML_TYPE_Q4_0, 128, 2,  64,  32, GGML_CUDA_MMQ_SRAM_LAYOUT_Q8_0, MMQ_ITER_K, false, true);
diff --git a/ggml/src/ggml-cuda/mmq-load-tiles.cuh b/ggml/src/ggml-cuda/mmq-load-tiles.cuh
index 7fb242096..8c9070c6b 100644
--- a/ggml/src/ggml-cuda/mmq-load-tiles.cuh
+++ b/ggml/src/ggml-cuda/mmq-load-tiles.cuh
@@ -95,6 +95,103 @@ template <ggml_type type, int J, bool fallback> static __device__ __forceinline_
     }
 }
 
+template <ggml_type type, int J, bool fallback> static __device__ __forceinline__ void ggml_cuda_mmq_load_tiles_q2_0(
+        const char * __restrict__ x, int * __restrict__ x_tile, const int kbx0, const int i_max, const int stride) {
+    constexpr int warp_size   = ggml_cuda_get_physical_warp_size();
+    constexpr int nwarps      = ggml_cuda_mmq_get_nthreads(type, J, fallback) / warp_size;
+    constexpr int I           = ggml_cuda_mmq_get_I(type, J, fallback);
+    constexpr int sram_stride = ggml_cuda_mmq_get_sram_stride(type, J, fallback);
+
+#if defined(AMD_MFMA_AVAILABLE) || defined(TURING_MMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
+    int   * x_qs = (int   *)  x_tile;
+    float * x_df = (float *) (x_qs + 2*MMQ_TILE_NE_K);
+#else
+    constexpr tile_x_sizes txs = mmq_get_dp4a_tile_x_sizes(GGML_TYPE_Q8_0, I);
+    int   * x_qs = (int   *)  x_tile;
+    float * x_df = (float *) (x_qs + txs.qs);
+#endif // defined(AMD_MFMA_AVAILABLE) || defined(TURING_MMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
+
+    constexpr int blocks_per_iter = MMQ_ITER_K / QK2_0;
+    constexpr int threads_per_row = blocks_per_iter * QI2_0;
+    constexpr int nrows = warp_size / threads_per_row;
+    constexpr int scale_entries_per_block = QK2_0 / QK8_1;
+    constexpr int scale_entries_per_row = blocks_per_iter * scale_entries_per_block;
+
+    const int txi  = threadIdx.x % threads_per_row;
+    const int kbx  = txi / QI2_0;
+    const int kqsx = txi % QI2_0;
+
+#pragma unroll
+    for (int i0 = 0; i0 < I; i0 += nrows*nwarps) {
+        int i = i0 + threadIdx.y*nrows + threadIdx.x/threads_per_row;
+
+        if (fallback) {
+            i = min(i, i_max);
+        }
+
+        const block_q2_0 * bxi = (const block_q2_0 *) x + kbx0 + i*stride + kbx;
+        // Each 32-element chunk occupies 8 bytes of qs (32 elements * 2 bits = 64 bits)
+        const int qs_offset = 8*kqsx;
+        const int qs0 = bxi->qs[qs_offset + 0] | (bxi->qs[qs_offset + 1] << 8) |
+                        (bxi->qs[qs_offset + 2] << 16) | (bxi->qs[qs_offset + 3] << 24);
+        const int qs1 = bxi->qs[qs_offset + 4] | (bxi->qs[qs_offset + 5] << 8) |
+                        (bxi->qs[qs_offset + 6] << 16) | (bxi->qs[qs_offset + 7] << 24);
+
+        // Unpack 32 2-bit codes into 8 int32s, each holding 4 signed int8s in {-1,0,1,2}.
+        int unpacked_bytes[8];
+#pragma unroll
+        for (int j = 0; j < 4; ++j) {
+            const int shift = j * 8;
+            const int codes = (qs0 >> shift) & 0xFF;
+            const int c0 = ((codes >> 0) & 0x3) - 1;
+            const int c1 = ((codes >> 2) & 0x3) - 1;
+            const int c2 = ((codes >> 4) & 0x3) - 1;
+            const int c3 = ((codes >> 6) & 0x3) - 1;
+            unpacked_bytes[j] = (c0 & 0xFF) | ((c1 & 0xFF) << 8) | ((c2 & 0xFF) << 16) | ((c3 & 0xFF) << 24);
+        }
+#pragma unroll
+        for (int j = 0; j < 4; ++j) {
+            const int shift = j * 8;
+            const int codes = (qs1 >> shift) & 0xFF;
+            const int c0 = ((codes >> 0) & 0x3) - 1;
+            const int c1 = ((codes >> 2) & 0x3) - 1;
+            const int c2 = ((codes >> 4) & 0x3) - 1;
+            const int c3 = ((codes >> 6) & 0x3) - 1;
+            unpacked_bytes[4 + j] = (c0 & 0xFF) | ((c1 & 0xFF) << 8) | ((c2 & 0xFF) << 16) | ((c3 & 0xFF) << 24);
+        }
+
+        const int dst_offset = kbx*(scale_entries_per_block*QI8_0) + kqsx*QI8_0;
+#pragma unroll
+        for (int j = 0; j < 8; ++j) {
+#if defined(AMD_MFMA_AVAILABLE) || defined(TURING_MMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
+            x_qs[i*sram_stride           + dst_offset + j] = unpacked_bytes[j];
+#else
+            x_qs[i*(2*MMQ_TILE_NE_K + 1) + dst_offset + j] = unpacked_bytes[j];
+#endif // defined(AMD_MFMA_AVAILABLE) || defined(TURING_MMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
+        }
+    }
+
+    const int ksx = threadIdx.x % scale_entries_per_row;
+    const int scale_block = ksx / scale_entries_per_block;
+
+#pragma unroll
+    for (int i0 = 0; i0 < I; i0 += nwarps) {
+        int i = i0 + threadIdx.y;
+
+        if (fallback) {
+            i = min(i, i_max);
+        }
+
+        const block_q2_0 * bxi = (const block_q2_0 *) x + kbx0 + i*stride + scale_block;
+
+#if defined(AMD_MFMA_AVAILABLE) || defined(TURING_MMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
+        x_df[i*sram_stride                           + ksx] = bxi->d;
+#else
+        x_df[i*(2*MMQ_TILE_NE_K/QI8_0) + i/(QI8_0/2) + ksx] = bxi->d;
+#endif // defined(AMD_MFMA_AVAILABLE) || defined(TURING_MMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
+    }
+}
+
 template <ggml_type type, int J, bool fallback> static __device__ __forceinline__ void ggml_cuda_mmq_load_tiles_q4_0(
         const char * __restrict__ x, int * __restrict__ x_tile, const int kbx0, const int i_max, const int stride) {
     constexpr int warp_size   = ggml_cuda_get_physical_warp_size();
diff --git a/ggml/src/ggml-cuda/mmq.cu b/ggml/src/ggml-cuda/mmq.cu
index 8cf1a3eb9..707437ea3 100644
--- a/ggml/src/ggml-cuda/mmq.cu
+++ b/ggml/src/ggml-cuda/mmq.cu
@@ -10,6 +10,9 @@ static void ggml_cuda_mul_mat_q_switch_type(ggml_backend_cuda_context & ctx, con
         case GGML_TYPE_Q1_0:
             mul_mat_q_case<GGML_TYPE_Q1_0>(ctx, args, stream);
             break;
+        case GGML_TYPE_Q2_0:
+            mul_mat_q_case<GGML_TYPE_Q2_0>(ctx, args, stream);
+            break;
         case GGML_TYPE_Q4_0:
             mul_mat_q_case<GGML_TYPE_Q4_0>(ctx, args, stream);
             break;
@@ -262,6 +265,7 @@ bool ggml_cuda_should_use_mmq(enum ggml_type type, int cc, int64_t ne11, int64_t
 
     switch (type) {
         case GGML_TYPE_Q1_0:
+        case GGML_TYPE_Q2_0:
         case GGML_TYPE_Q4_0:
         case GGML_TYPE_Q4_1:
         case GGML_TYPE_Q5_0:
diff --git a/ggml/src/ggml-cuda/mmq.cuh b/ggml/src/ggml-cuda/mmq.cuh
index 5f30f5f6b..2eb15fdfa 100644
--- a/ggml/src/ggml-cuda/mmq.cuh
+++ b/ggml/src/ggml-cuda/mmq.cuh
@@ -60,6 +60,7 @@ static_assert(sizeof(block_fp4_mmq)  == sizeof(block_q8_1_mmq),    "Unexpected b
 static mmq_q8_1_ds_layout mmq_get_q8_1_ds_layout(const ggml_type type_x) {
     switch (type_x) {
         case GGML_TYPE_Q1_0:
+        case GGML_TYPE_Q2_0:
             return MMQ_Q8_1_DS_LAYOUT_D4;
         case GGML_TYPE_Q4_0:
         case GGML_TYPE_Q4_1:
@@ -385,6 +386,7 @@ static constexpr __device__ int ggml_cuda_mmq_get_rows_per_warp(ggml_type type,
 static constexpr __host__ __device__ tile_x_sizes mmq_get_dp4a_tile_x_sizes(ggml_type type, int I) {
     switch (type) {
         case GGML_TYPE_Q1_0:    return MMQ_DP4A_TXS_Q8_0;
+        case GGML_TYPE_Q2_0:    return MMQ_DP4A_TXS_Q8_0;
         case GGML_TYPE_Q4_0:    return MMQ_DP4A_TXS_Q4_0;
         case GGML_TYPE_Q4_1:    return MMQ_DP4A_TXS_Q4_1;
         case GGML_TYPE_Q5_0:    return MMQ_DP4A_TXS_Q8_0;
@@ -542,6 +544,12 @@ static constexpr __device__ ggml_cuda_mmq_util_funcs ggml_cuda_mmq_get_util_func
                     ggml_cuda_mmq_load_tiles_q1_0<type, J, fallback>,
                     ggml_cuda_mmq_vec_dot_q8_0_q8_1_dp4a<type, J, fallback>,
                     ggml_cuda_mmq_write_back_dp4a<type, J, fallback>);
+            case GGML_TYPE_Q2_0:
+                return ggml_cuda_mmq_util_funcs(
+                    VDR_Q2_0_Q8_1_MMQ,
+                    ggml_cuda_mmq_load_tiles_q2_0<type, J, fallback>,
+                    ggml_cuda_mmq_vec_dot_q8_0_q8_1_dp4a<type, J, fallback>,
+                    ggml_cuda_mmq_write_back_dp4a<type, J, fallback>);
             case GGML_TYPE_Q4_0:
                 return ggml_cuda_mmq_util_funcs(
                     VDR_Q4_0_Q8_1_MMQ,
@@ -700,6 +708,12 @@ static constexpr __device__ ggml_cuda_mmq_util_funcs ggml_cuda_mmq_get_util_func
                 ggml_cuda_mmq_load_tiles_q1_0<type, J, fallback>,
                 ggml_cuda_mmq_vec_dot_q8_0_q8_1_mma<type, J, fallback, MMQ_Q8_1_DS_LAYOUT_D4>,
                 ggml_cuda_mmq_write_back_mma<type, J, fallback>);
+        case GGML_TYPE_Q2_0:
+            return ggml_cuda_mmq_util_funcs(
+                -1,
+                ggml_cuda_mmq_load_tiles_q2_0<type, J, fallback>,
+                ggml_cuda_mmq_vec_dot_q8_0_q8_1_mma<type, J, fallback, MMQ_Q8_1_DS_LAYOUT_D4>,
+                ggml_cuda_mmq_write_back_mma<type, J, fallback>);
         case GGML_TYPE_Q4_0:
             return ggml_cuda_mmq_util_funcs(
                 -1,
@@ -1550,6 +1564,7 @@ void mul_mat_q_case(ggml_backend_cuda_context & ctx, const mmq_args & args, cuda
     template void mul_mat_q_case<type>(ggml_backend_cuda_context & ctx, const mmq_args & args, cudaStream_t stream) \
 
 extern DECL_MMQ_CASE(GGML_TYPE_Q1_0);
+extern DECL_MMQ_CASE(GGML_TYPE_Q2_0);
 extern DECL_MMQ_CASE(GGML_TYPE_Q4_0);
 extern DECL_MMQ_CASE(GGML_TYPE_Q4_1);
 extern DECL_MMQ_CASE(GGML_TYPE_Q5_0);
diff --git a/ggml/src/ggml-cuda/mmvq.cu b/ggml/src/ggml-cuda/mmvq.cu
index e18ada537..0589e65bd 100644
--- a/ggml/src/ggml-cuda/mmvq.cu
+++ b/ggml/src/ggml-cuda/mmvq.cu
@@ -10,6 +10,7 @@ typedef float (*vec_dot_q_cuda_t)(const void * __restrict__ vbq, const block_q8_
 static constexpr __device__ vec_dot_q_cuda_t get_vec_dot_q_cuda(ggml_type type) {
     switch (type) {
         case GGML_TYPE_Q1_0:    return vec_dot_q1_0_q8_1;
+        case GGML_TYPE_Q2_0:    return vec_dot_q2_0_q8_1;
         case GGML_TYPE_Q4_0:    return vec_dot_q4_0_q8_1;
         case GGML_TYPE_Q4_1:    return vec_dot_q4_1_q8_1;
         case GGML_TYPE_Q5_0:    return vec_dot_q5_0_q8_1;
@@ -38,6 +39,7 @@ static constexpr __device__ vec_dot_q_cuda_t get_vec_dot_q_cuda(ggml_type type)
 static constexpr __host__ __device__ int get_vdr_mmvq(ggml_type type) {
     switch (type) {
         case GGML_TYPE_Q1_0:    return VDR_Q1_0_Q8_1_MMVQ;
+        case GGML_TYPE_Q2_0:    return VDR_Q2_0_Q8_1_MMVQ;
         case GGML_TYPE_Q4_0:    return VDR_Q4_0_Q8_1_MMVQ;
         case GGML_TYPE_Q4_1:    return VDR_Q4_1_Q8_1_MMVQ;
         case GGML_TYPE_Q5_0:    return VDR_Q5_0_Q8_1_MMVQ;
@@ -1010,6 +1012,12 @@ static void mul_mat_vec_q_switch_type(
                  nchannels_x, nchannels_y, nchannels_dst, stride_channel_x, stride_channel_y, stride_channel_dst,
                  nsamples_x, nsamples_dst, stride_sample_x, stride_sample_y, stride_sample_dst, ids_stride, stream);
             break;
+        case GGML_TYPE_Q2_0:
+            mul_mat_vec_q_switch_ncols_dst<GGML_TYPE_Q2_0>
+                (vx, vy, ids, fusion, dst, ncols_x, nrows_x, ncols_dst, stride_row_x, stride_col_y, stride_col_dst,
+                 nchannels_x, nchannels_y, nchannels_dst, stride_channel_x, stride_channel_y, stride_channel_dst,
+                 nsamples_x, nsamples_dst, stride_sample_x, stride_sample_y, stride_sample_dst, ids_stride, stream);
+            break;
         case GGML_TYPE_Q4_0:
             mul_mat_vec_q_switch_ncols_dst<GGML_TYPE_Q4_0>
                 (vx, vy, ids, fusion, dst, ncols_x, nrows_x, ncols_dst, stride_row_x, stride_col_y, stride_col_dst,
diff --git a/ggml/src/ggml-cuda/template-instances/generate_cu_files.py b/ggml/src/ggml-cuda/template-instances/generate_cu_files.py
index 614b1566c..d7cd27167 100755
--- a/ggml/src/ggml-cuda/template-instances/generate_cu_files.py
+++ b/ggml/src/ggml-cuda/template-instances/generate_cu_files.py
@@ -36,6 +36,7 @@ SOURCE_FATTN_MMA_CASE = "DECL_FATTN_MMA_F16_CASE({head_size_kq}, {head_size_v},
 
 TYPES_MMQ = [
     "GGML_TYPE_Q1_0",
+    "GGML_TYPE_Q2_0",
     "GGML_TYPE_Q4_0", "GGML_TYPE_Q4_1", "GGML_TYPE_Q5_0", "GGML_TYPE_Q5_1", "GGML_TYPE_Q8_0",
     "GGML_TYPE_Q2_K", "GGML_TYPE_Q3_K", "GGML_TYPE_Q4_K", "GGML_TYPE_Q5_K", "GGML_TYPE_Q6_K",
     "GGML_TYPE_IQ2_XXS", "GGML_TYPE_IQ2_XS", "GGML_TYPE_IQ2_S", "GGML_TYPE_IQ3_XXS", "GGML_TYPE_IQ3_S",
diff --git a/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_0.cu b/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_0.cu
new file mode 100644
index 000000000..750180e33
--- /dev/null
+++ b/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_0.cu
@@ -0,0 +1,5 @@
+// This file has been autogenerated by generate_cu_files.py, do not edit manually.
+
+#include "../mmq.cuh"
+
+DECL_MMQ_CASE(GGML_TYPE_Q2_0);
diff --git a/ggml/src/ggml-cuda/vecdotq.cuh b/ggml/src/ggml-cuda/vecdotq.cuh
index b9932bce9..36d89c89e 100644
--- a/ggml/src/ggml-cuda/vecdotq.cuh
+++ b/ggml/src/ggml-cuda/vecdotq.cuh
@@ -109,6 +109,9 @@ static __device__ __forceinline__ uint32_t unpack_ksigns(const uint8_t v) {
 #define VDR_Q1_0_Q8_1_MMVQ 1  // Process one 32-element chunk at a time for parallelism
 #define VDR_Q1_0_Q8_1_MMQ  4  // Q1_0 has 128 bits (4 ints) per block
 
+#define VDR_Q2_0_Q8_1_MMVQ 1  // Process one 32-element chunk at a time for parallelism
+#define VDR_Q2_0_Q8_1_MMQ  2  // Q2_0 group 64: 128 bits (4 ints) per block, 2 32-element chunks
+
 #define VDR_Q4_0_Q8_1_MMVQ 2
 #define VDR_Q4_0_Q8_1_MMQ  4
 
@@ -722,6 +725,64 @@ static __device__ __forceinline__ float vec_dot_q1_0_q8_1(
     return d1 * d8 * sumi;
 }
 
+static __device__ __forceinline__ float vec_dot_q2_0_q8_1(
+    const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) {
+
+    const block_q2_0 * bq2_0 = (const block_q2_0 *) vbq + kbx;
+
+    // Q2_0 (group 64): 64 elements with ONE scale, 2 bits per element (4 elements per byte)
+    // Q8_1: 32 elements per block with individual scales
+    // iqs selects which of the 2 chunks of 32 elements to process (0-1)
+
+    const float d2 = bq2_0->d;
+
+    // Process only the chunk specified by iqs
+    const block_q8_1 * bq8_1_chunk = bq8_1 + iqs;
+
+    // Load 64 bits (8 bytes) for this chunk from Q2_0: bytes [8*iqs, 8*iqs+8)
+    const int offset = iqs * 8;
+    const int v0 = bq2_0->qs[offset + 0] | (bq2_0->qs[offset + 1] << 8) |
+                   (bq2_0->qs[offset + 2] << 16) | (bq2_0->qs[offset + 3] << 24);
+    const int v1 = bq2_0->qs[offset + 4] | (bq2_0->qs[offset + 5] << 8) |
+                   (bq2_0->qs[offset + 6] << 16) | (bq2_0->qs[offset + 7] << 24);
+
+    // Unpack 32 2-bit codes into 8 int32s, each holding 4 signed int8 symbols in {-1,0,1,2}.
+    // Stored code c in {0,1,2,3} -> symbol s = c - 1.
+    int vi_bytes[8];
+#pragma unroll
+    for (int j = 0; j < 4; ++j) {
+        const int shift = j * 8;
+        const int codes = (v0 >> shift) & 0xFF;
+        const int c0 = ((codes >> 0) & 0x3) - 1;
+        const int c1 = ((codes >> 2) & 0x3) - 1;
+        const int c2 = ((codes >> 4) & 0x3) - 1;
+        const int c3 = ((codes >> 6) & 0x3) - 1;
+        vi_bytes[j] = (c0 & 0xFF) | ((c1 & 0xFF) << 8) | ((c2 & 0xFF) << 16) | ((c3 & 0xFF) << 24);
+    }
+#pragma unroll
+    for (int j = 0; j < 4; ++j) {
+        const int shift = j * 8;
+        const int codes = (v1 >> shift) & 0xFF;
+        const int c0 = ((codes >> 0) & 0x3) - 1;
+        const int c1 = ((codes >> 2) & 0x3) - 1;
+        const int c2 = ((codes >> 4) & 0x3) - 1;
+        const int c3 = ((codes >> 6) & 0x3) - 1;
+        vi_bytes[4 + j] = (c0 & 0xFF) | ((c1 & 0xFF) << 8) | ((c2 & 0xFF) << 16) | ((c3 & 0xFF) << 24);
+    }
+
+    // Compute dot product for this 32-element chunk
+    int sumi = 0;
+#pragma unroll
+    for (int j = 0; j < 8; ++j) {
+        const int u = get_int_b4(bq8_1_chunk->qs, j);
+        sumi = ggml_cuda_dp4a(vi_bytes[j], u, sumi);
+    }
+
+    // Apply Q2_0's single scale and this chunk's Q8_1 scale
+    const float d8 = __low2float(bq8_1_chunk->ds);
+    return d2 * d8 * sumi;
+}
+
 static __device__ __forceinline__ float vec_dot_q4_0_q8_1(
     const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) {
 
diff --git a/tests/test-backend-ops.cpp b/tests/test-backend-ops.cpp
index d0eb173a8..ef18ecef8 100644
--- a/tests/test-backend-ops.cpp
+++ b/tests/test-backend-ops.cpp
@@ -8013,6 +8013,7 @@ static const ggml_type other_types[] = {
     GGML_TYPE_Q5_0, GGML_TYPE_Q5_1,
     GGML_TYPE_Q8_0,
     GGML_TYPE_Q1_0,
+    GGML_TYPE_Q2_0,
     GGML_TYPE_Q2_K, GGML_TYPE_Q3_K,
     GGML_TYPE_Q5_K,
     GGML_TYPE_Q6_K,
@@ -9563,6 +9564,10 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
     test_cases.emplace_back(new test_flash_attn_ext(128, 64, 4, {1, 1}, 128, 2, true, false, 0, 0, GGML_PREC_F32, GGML_TYPE_Q1_0, GGML_TYPE_Q4_0));
     test_cases.emplace_back(new test_flash_attn_ext(64, 128, 4, {1, 1}, 128, 2, true, false, 0, 0, GGML_PREC_F32, GGML_TYPE_Q4_0, GGML_TYPE_Q1_0));
     test_cases.emplace_back(new test_flash_attn_ext(128, 64, 4, {1, 1}, 64, 2, true, false, 0, 0, GGML_PREC_F32, GGML_TYPE_Q1_0, GGML_TYPE_F16));
+    test_cases.emplace_back(new test_flash_attn_ext(128, 128, 4, {1, 1}, 96, 2, true, false, 0, 0, GGML_PREC_F32, GGML_TYPE_Q2_0, GGML_TYPE_Q2_0));
+    test_cases.emplace_back(new test_flash_attn_ext(128, 64, 4, {1, 1}, 128, 2, true, false, 0, 0, GGML_PREC_F32, GGML_TYPE_Q2_0, GGML_TYPE_Q4_0));
+    test_cases.emplace_back(new test_flash_attn_ext(64, 128, 4, {1, 1}, 128, 2, true, false, 0, 0, GGML_PREC_F32, GGML_TYPE_Q4_0, GGML_TYPE_Q2_0));
+    test_cases.emplace_back(new test_flash_attn_ext(128, 64, 4, {1, 1}, 64, 2, true, false, 0, 0, GGML_PREC_F32, GGML_TYPE_Q2_0, GGML_TYPE_F16));
 
     // large-KV F16 cases (Qwen3.6-27B geometry and a llama-class control): the upstream matrix
     // stops at kv=1024, blind to long-context FA bugs (e.g. the oneDNN SDPA ordering race on BMG).

From 6a4c3357c86517f2ae0a853436e130672a67147c Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 30 Jul 2026 15:42:47 +0300
Subject: [PATCH 11/30] ggml : bump version to 0.18.0 (ggml/1576)

---
 ggml/CMakeLists.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ggml/CMakeLists.txt b/ggml/CMakeLists.txt
index a766e49ea..159da3afa 100644
--- a/ggml/CMakeLists.txt
+++ b/ggml/CMakeLists.txt
@@ -4,7 +4,7 @@ project("ggml" C CXX ASM)
 
 ### GGML Version
 set(GGML_VERSION_MAJOR 0)
-set(GGML_VERSION_MINOR 17)
+set(GGML_VERSION_MINOR 18)
 set(GGML_VERSION_PATCH 0)
 set(GGML_VERSION_BASE "${GGML_VERSION_MAJOR}.${GGML_VERSION_MINOR}.${GGML_VERSION_PATCH}")
 

From 9ebfc3a8cf4c1c6983258c4d603274b2b3d3dd65 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 30 Jul 2026 15:43:44 +0300
Subject: [PATCH 12/30] sync : ggml

---
 scripts/sync-ggml.last | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/sync-ggml.last b/scripts/sync-ggml.last
index 24e27e6f0..230e59a95 100644
--- a/scripts/sync-ggml.last
+++ b/scripts/sync-ggml.last
@@ -1 +1 @@
-9be313313c8ecb9488911bd64550190e3ed80f38
+06ca97616793248fadb410ea8d69c7511b2005e4

From 6b36c2305644fd30db7cce3f4840c74574f31ce9 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 30 Jul 2026 16:14:37 +0300
Subject: [PATCH 13/30] readme : refresh (#26280)

* docs : center badges and links, remove Hot topics

- Use <div align="center"> for GitHub-compatible centering
- Add dev branches and compile times links
- Remove Hot topics section

Assisted-by: llama.cpp:Qwen3.6-27B

* readme : remove sections

* docs : center badges, remove Hot topics, extract sections, remove tools

- Use <div align="center"> for GitHub-compatible centering
- Add dev branches and compile times links
- Add lib llama API and llama-server REST API links
- Remove Hot topics section
- Remove Recent API changes section
- Extract XCFramework section into docs/xcframework.md
- Extract Completions section into docs/completions.md
- Extract Obtaining and quantizing models into docs/models.md
- Remove tools usage sections (llama-cli, llama-server, etc.)
- Move Contributing section to the end

Assisted-by: llama.cpp:Qwen3.6-27B

* cont : arrange links

* cont : fix ws

* cont : remove seminal papers

* cont : change sample model

* cont : trim-down contributing section

* cont : sort backends alphabetically

* cont : words

* cont : add fig captions

* docs : models words

* readme : shorter caption

* cont : fix typo

* cont : add window frame to screenshot
---
 README.md           | 584 ++++----------------------------------------
 docs/completions.md |  17 ++
 docs/models.md      |  26 ++
 docs/xcframework.md |  31 +++
 4 files changed, 126 insertions(+), 532 deletions(-)
 create mode 100644 docs/completions.md
 create mode 100644 docs/models.md
 create mode 100644 docs/xcframework.md

diff --git a/README.md b/README.md
index ee70581e3..57436327e 100644
--- a/README.md
+++ b/README.md
@@ -2,65 +2,56 @@
 
 ![llama](https://raw.githubusercontent.com/ggml-org/llama.brand/refs/heads/master/cover/llama-cpp/cover-llama-cpp-dark.svg)
 
+<div align="center">
+
+<b>LLM inference in C/C++</b>
+
 [![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](https://opensource.org/licenses/MIT)
 [![Release](https://img.shields.io/github/v/release/ggml-org/llama.cpp)](https://github.com/ggml-org/llama.cpp/releases)
 [![Server](https://github.com/ggml-org/llama.cpp/actions/workflows/server.yml/badge.svg)](https://github.com/ggml-org/llama.cpp/actions/workflows/server.yml)
 [![Docker](https://github.com/ggml-org/llama.cpp/actions/workflows/docker.yml/badge.svg)](https://github.com/ggml-org/llama.cpp/actions/workflows/docker.yml)
 [![Winget](https://github.com/ggml-org/llama.cpp/actions/workflows/winget.yml/badge.svg)](https://github.com/ggml-org/llama.cpp/actions/workflows/winget.yml)
 
-[Manifesto](https://github.com/ggml-org/llama.cpp/discussions/205) / [ggml](https://github.com/ggml-org/ggml) / [ops](https://github.com/ggml-org/llama.cpp/blob/master/docs/ops.md) / [maintainer PRs](https://github.com/ggml-org/llama.cpp/issues?q=is%3Apr%20is%3Aopen%20draft%3AFalse%20(author%3Argerganov%20OR%20author%3AKitaitiMakoto%20OR%20author%3Adanbev%20OR%20author%3Aaldehir%20OR%20author%3Amax-krasnyansky%20OR%20author%3ACISC%20OR%20author%3Aggerganov%20OR%20author%3Aam17an%20OR%20author%3Abartowski1182%20OR%20author%3Ahipudding%20OR%20author%3AServeurpersoCom%20OR%20author%3Apwilkin%20OR%20author%3Areeselevine%20OR%20author%3Angxson%20OR%20author%3Ajeffbolznv%20OR%20author%3A0cc4m%20OR%20author%3Aangt%20OR%20author%3AIMbackK%20OR%20author%3Aarthw%20OR%20author%3AJohannesGaessler%20OR%20author%3AORippler%20OR%20author%3Aruixiang63%20OR%20author%3Axctan%20OR%20author%3Aallozaur%20OR%20author%3Ayomaytk%20OR%20author%3Aaendk%20OR%20author%3Agaugarg-nv%20OR%20author%3Ataronaeo%20OR%20author%3Aforforever73%20OR%20author%3Alhez%20OR%20author%3Anetrunnereve%20OR%20author%3Afairydreaming)%20sort%3Aupdated-desc)
+[manifesto](https://github.com/ggml-org/llama.cpp/discussions/205) / [ggml](https://github.com/ggml-org/ggml) / [ops](https://github.com/ggml-org/llama.cpp/blob/master/docs/ops.md) / [maintainer PRs](https://github.com/ggml-org/llama.cpp/issues?q=is%3Apr%20is%3Aopen%20draft%3AFalse%20(author%3Argerganov%20OR%20author%3AKitaitiMakoto%20OR%20author%3Adanbev%20OR%20author%3Aaldehir%20OR%20author%3Amax-krasnyansky%20OR%20author%3ACISC%20OR%20author%3Aggerganov%20OR%20author%3Aam17an%20OR%20author%3Abartowski1182%20OR%20author%3Ahipudding%20OR%20author%3AServeurpersoCom%20OR%20author%3Apwilkin%20OR%20author%3Areeselevine%20OR%20author%3Angxson%20OR%20author%3Ajeffbolznv%20OR%20author%3A0cc4m%20OR%20author%3Aangt%20OR%20author%3AIMbackK%20OR%20author%3Aarthw%20OR%20author%3AJohannesGaessler%20OR%20author%3AORippler%20OR%20author%3Aruixiang63%20OR%20author%3Axctan%20OR%20author%3Aallozaur%20OR%20author%3Ayomaytk%20OR%20author%3Aaendk%20OR%20author%3Agaugarg-nv%20OR%20author%3Ataronaeo%20OR%20author%3Aforforever73%20OR%20author%3Alhez%20OR%20author%3Anetrunnereve%20OR%20author%3Afairydreaming)%20sort%3Aupdated-desc) / [dev branches](https://github.com/ggml-org/llama.cpp-dev/blob/master/README-features.md) / [compile times](https://github.com/ggml-org/llama.cpp-dev/blob/master/README-compile-times.md) / [lib llama API](https://github.com/ggml-org/llama.cpp/issues/9289) / [llama-server REST API](https://github.com/ggml-org/llama.cpp/issues/9291)
 
-LLM inference in C/C++
-
-## Recent API changes
-
-- [Changelog for `libllama` API](https://github.com/ggml-org/llama.cpp/issues/9289)
-- [Changelog for `llama-server` REST API](https://github.com/ggml-org/llama.cpp/issues/9291)
-
-## Hot topics
-
-- **Hugging Face cache migration: models downloaded with `-hf` are now stored in the standard Hugging Face cache directory, enabling sharing with other HF tools.**
-- **[guide : using the new WebUI of llama.cpp](https://github.com/ggml-org/llama.cpp/discussions/16938)**
-- [guide : running gpt-oss with llama.cpp](https://github.com/ggml-org/llama.cpp/discussions/15396)
-- [[FEEDBACK] Better packaging for llama.cpp to support downstream consumers 🤗](https://github.com/ggml-org/llama.cpp/discussions/15313)
-- Support for the `gpt-oss` model with native MXFP4 format has been added | [PR](https://github.com/ggml-org/llama.cpp/pull/15091) | [Collaboration with NVIDIA](https://blogs.nvidia.com/blog/rtx-ai-garage-openai-oss) | [Comment](https://github.com/ggml-org/llama.cpp/discussions/15095)
-- Multimodal support arrived in `llama-server`: [#12898](https://github.com/ggml-org/llama.cpp/pull/12898) | [documentation](./docs/multimodal.md)
-- VS Code extension for FIM completions: https://github.com/ggml-org/llama.vscode
-- Vim/Neovim plugin for FIM completions: https://github.com/ggml-org/llama.vim
-- Hugging Face Inference Endpoints now support GGUF out of the box! https://github.com/ggml-org/llama.cpp/discussions/9669
-- Hugging Face GGUF editor: [discussion](https://github.com/ggml-org/llama.cpp/discussions/9268) | [tool](https://huggingface.co/spaces/CISCai/gguf-editor)
-- WebGPU support is now available in the browser, see a blog/demo introducing it [here](https://reeselevine.github.io/llamas-on-the-web/).
-
-----
+</div>
 
 ## Quick start
 
-Getting started with llama.cpp is straightforward. Here are several ways to install it on your machine:
+A few options to get `llama.cpp` installed on your machine:
 
-- Install `llama.cpp` using [brew, nix, winget, or conda-forge](docs/install.md)
+- Visit https://llama.app and follow the instructions
 - Run with Docker - see our [Docker documentation](docs/docker.md)
 - Download pre-built binaries from the [releases page](https://github.com/ggml-org/llama.cpp/releases)
 - Build from source by cloning this repository - check out [our build guide](docs/build.md)
 
-Once installed, you'll need a model to work with. Head to the [Obtaining and quantizing models](#obtaining-and-quantizing-models) section to learn more.
-
-Example command:
+Once installed:
 
 ```sh
-# Use a local model file
-llama-cli -m my_model.gguf
-
-# Or download and run a model directly from Hugging Face
-llama-cli -hf ggml-org/gemma-3-1b-it-GGUF
+# Download and run a model directly from Hugging Face
+llama cli -hf ggml-org/Qwen3.5-0.8B-GGUF
 
 # Launch OpenAI-compatible API server
-llama-server -hf ggml-org/gemma-3-1b-it-GGUF
+llama serve -hf ggml-org/Qwen3.5-0.8B-GGUF
 ```
 
+<table align="center">
+    <tr>
+        <td align="center" width=50%>
+            <img width="1310" height="888" alt="VLM session with `llama cli`" src="https://github.com/user-attachments/assets/88726b48-1713-48aa-a525-95a02e78afc4" />
+            <i>VLM session with <b>llama cli</b></i>
+        </td>
+        <td align="center">
+            <img width="1392" height="958" alt="Built-in web UI against `llama serve` running Qwen 3.6" src="https://github.com/user-attachments/assets/b402f972-2e32-4def-8771-8d849f08cf2e" />
+            <i>Built-in web UI against <b>llama serve</b></i>
+        </td>
+    </tr>
+<table>
+
 ## Description
 
-The main goal of `llama.cpp` is to enable LLM inference with minimal setup and state-of-the-art performance on a wide
-range of hardware - locally and in the cloud.
+The main goal of `llama.cpp` is to enable LLM (and VLM) inference with minimal setup and state-of-the-art performance on
+a wide range of hardware - locally and in the cloud.
 
 - Plain C/C++ implementation without any dependencies
 - Apple silicon is a first-class citizen - optimized via ARM NEON, Accelerate and Metal frameworks
@@ -71,467 +62,40 @@ range of hardware - locally and in the cloud.
 - Vulkan and SYCL backend support
 - CPU+GPU hybrid inference to partially accelerate models larger than the total VRAM capacity
 
-The `llama.cpp` project is the main playground for developing new features for the [ggml](https://github.com/ggml-org/ggml) library.
-
-<details>
-<summary>Models</summary>
-
-Typically finetunes of the base models below are supported as well.
-
-Instructions for adding support for new models: [HOWTO-add-model.md](docs/development/HOWTO-add-model.md)
-
-#### Text-only
-
-- [X] LLaMA 🦙
-- [x] LLaMA 2 🦙🦙
-- [x] LLaMA 3 🦙🦙🦙
-- [X] [Mistral 7B](https://huggingface.co/mistralai/Mistral-7B-v0.1)
-- [x] [Mixtral MoE](https://huggingface.co/models?search=mistral-ai/Mixtral)
-- [x] [DBRX](https://huggingface.co/databricks/dbrx-instruct)
-- [x] [Jamba](https://huggingface.co/ai21labs)
-- [X] [Falcon](https://huggingface.co/models?search=tiiuae/falcon)
-- [X] [Chinese LLaMA / Alpaca](https://github.com/ymcui/Chinese-LLaMA-Alpaca) and [Chinese LLaMA-2 / Alpaca-2](https://github.com/ymcui/Chinese-LLaMA-Alpaca-2)
-- [X] [Vigogne (French)](https://github.com/bofenghuang/vigogne)
-- [X] [BERT](https://github.com/ggml-org/llama.cpp/pull/5423)
-- [X] [Koala](https://bair.berkeley.edu/blog/2023/04/03/koala/)
-- [X] [Baichuan 1 & 2](https://huggingface.co/models?search=baichuan-inc/Baichuan) + [derivations](https://huggingface.co/hiyouga/baichuan-7b-sft)
-- [X] [Aquila 1 & 2](https://huggingface.co/models?search=BAAI/Aquila)
-- [X] [Starcoder models](https://github.com/ggml-org/llama.cpp/pull/3187)
-- [X] [Refact](https://huggingface.co/smallcloudai/Refact-1_6B-fim)
-- [X] [MPT](https://github.com/ggml-org/llama.cpp/pull/3417)
-- [X] [Bloom](https://github.com/ggml-org/llama.cpp/pull/3553)
-- [x] [Yi models](https://huggingface.co/models?search=01-ai/Yi)
-- [X] [StableLM models](https://huggingface.co/stabilityai)
-- [x] [Deepseek models](https://huggingface.co/models?search=deepseek-ai/deepseek)
-- [x] [Qwen models](https://huggingface.co/models?search=Qwen/Qwen)
-- [x] [PLaMo-13B](https://github.com/ggml-org/llama.cpp/pull/3557)
-- [x] [Phi models](https://huggingface.co/models?search=microsoft/phi)
-- [x] [PhiMoE](https://github.com/ggml-org/llama.cpp/pull/11003)
-- [x] [GPT-2](https://huggingface.co/gpt2)
-- [x] [Orion 14B](https://github.com/ggml-org/llama.cpp/pull/5118)
-- [x] [InternLM2](https://huggingface.co/models?search=internlm2)
-- [x] [CodeShell](https://github.com/WisdomShell/codeshell)
-- [x] [Gemma](https://ai.google.dev/gemma)
-- [x] [Mamba](https://github.com/state-spaces/mamba)
-- [x] [Grok-1](https://huggingface.co/keyfan/grok-1-hf)
-- [x] [Xverse](https://huggingface.co/models?search=xverse)
-- [x] [Command-R models](https://huggingface.co/models?search=CohereForAI/c4ai-command-r)
-- [x] [SEA-LION](https://huggingface.co/models?search=sea-lion)
-- [x] [GritLM-7B](https://huggingface.co/GritLM/GritLM-7B) + [GritLM-8x7B](https://huggingface.co/GritLM/GritLM-8x7B)
-- [x] [OLMo](https://allenai.org/olmo)
-- [x] [OLMo 2](https://allenai.org/olmo)
-- [x] [OLMoE](https://huggingface.co/allenai/OLMoE-1B-7B-0924)
-- [x] [Granite models](https://huggingface.co/collections/ibm-granite/granite-code-models-6624c5cec322e4c148c8b330)
-- [x] [GPT-NeoX](https://github.com/EleutherAI/gpt-neox) + [Pythia](https://github.com/EleutherAI/pythia)
-- [x] [Snowflake-Arctic MoE](https://huggingface.co/collections/Snowflake/arctic-66290090abe542894a5ac520)
-- [x] [Smaug](https://huggingface.co/models?search=Smaug)
-- [x] [Poro 34B](https://huggingface.co/LumiOpen/Poro-34B)
-- [x] [Bitnet b1.58 models](https://huggingface.co/1bitLLM)
-- [x] [Flan T5](https://huggingface.co/models?search=flan-t5)
-- [x] [Open Elm models](https://huggingface.co/collections/apple/openelm-instruct-models-6619ad295d7ae9f868b759ca)
-- [x] [ChatGLM3-6b](https://huggingface.co/THUDM/chatglm3-6b) + [ChatGLM4-9b](https://huggingface.co/THUDM/glm-4-9b) + [GLMEdge-1.5b](https://huggingface.co/THUDM/glm-edge-1.5b-chat) + [GLMEdge-4b](https://huggingface.co/THUDM/glm-edge-4b-chat)
-- [x] [GLM-4-0414](https://huggingface.co/collections/THUDM/glm-4-0414-67f3cbcb34dd9d252707cb2e)
-- [x] [SmolLM](https://huggingface.co/collections/HuggingFaceTB/smollm-6695016cad7167254ce15966)
-- [x] [EXAONE-3.0-7.8B-Instruct](https://huggingface.co/LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct)
-- [x] [FalconMamba Models](https://huggingface.co/collections/tiiuae/falconmamba-7b-66b9a580324dd1598b0f6d4a)
-- [x] [Jais](https://huggingface.co/inceptionai/jais-13b-chat)
-- [x] [Bielik-11B-v2.3](https://huggingface.co/collections/speakleash/bielik-11b-v23-66ee813238d9b526a072408a)
-- [x] [RWKV-7](https://huggingface.co/collections/shoumenchougou/rwkv7-gxx-gguf)
-- [x] [RWKV-6](https://github.com/BlinkDL/RWKV-LM)
-- [x] [QRWKV-6](https://huggingface.co/recursal/QRWKV6-32B-Instruct-Preview-v0.1)
-- [x] [GigaChat-20B-A3B](https://huggingface.co/ai-sage/GigaChat-20B-A3B-instruct)
-- [X] [Trillion-7B-preview](https://huggingface.co/trillionlabs/Trillion-7B-preview)
-- [x] [Ling models](https://huggingface.co/collections/inclusionAI/ling-67c51c85b34a7ea0aba94c32)
-- [x] [Liquid LFM2 models](https://huggingface.co/collections/LiquidAI/lfm2)
-- [x] [Liquid LFM2.5 models](https://huggingface.co/collections/LiquidAI/lfm25)
-- [x] [Liquid Nanos](https://huggingface.co/collections/LiquidAI/liquid-nanos)
-- [x] [Hunyuan models](https://huggingface.co/collections/tencent/hunyuan-dense-model-6890632cda26b19119c9c5e7)
-- [x] [BailingMoeV2 (Ring/Ling 2.0) models](https://huggingface.co/collections/inclusionAI/ling-v2-68bf1dd2fc34c306c1fa6f86)
-- [x] [Mellum models](https://huggingface.co/JetBrains/models?search=mellum)
-
-#### Multimodal
-
-- [x] [LLaVA 1.5 models](https://huggingface.co/collections/liuhaotian/llava-15-653aac15d994e992e2677a7e), [LLaVA 1.6 models](https://huggingface.co/collections/liuhaotian/llava-16-65b9e40155f60fd046a5ccf2)
-- [x] [BakLLaVA](https://huggingface.co/models?search=SkunkworksAI/Bakllava)
-- [x] [Obsidian](https://huggingface.co/NousResearch/Obsidian-3B-V0.5)
-- [x] [ShareGPT4V](https://huggingface.co/models?search=Lin-Chen/ShareGPT4V)
-- [x] [MobileVLM 1.7B/3B models](https://huggingface.co/models?search=mobileVLM)
-- [x] [Yi-VL](https://huggingface.co/models?search=Yi-VL)
-- [x] [Mini CPM](https://huggingface.co/models?search=MiniCPM)
-- [x] [Moondream](https://huggingface.co/vikhyatk/moondream2)
-- [x] [Bunny](https://github.com/BAAI-DCAI/Bunny)
-- [x] [GLM-EDGE](https://huggingface.co/models?search=glm-edge)
-- [x] [Qwen2-VL](https://huggingface.co/collections/Qwen/qwen2-vl-66cee7455501d7126940800d)
-- [x] [LFM2-VL](https://huggingface.co/collections/LiquidAI/lfm2-vl-68963bbc84a610f7638d5ffa)
-
-</details>
-
-<details>
-<summary>Bindings</summary>
-
-- Python: [ddh0/easy-llama](https://github.com/ddh0/easy-llama)
-- Python: [abetlen/llama-cpp-python](https://github.com/abetlen/llama-cpp-python)
-- Go: [go-skynet/go-llama.cpp](https://github.com/go-skynet/go-llama.cpp)
-- Node.js: [withcatai/node-llama-cpp](https://github.com/withcatai/node-llama-cpp)
-- JS/TS (llama.cpp server client): [lgrammel/modelfusion](https://modelfusion.dev/integration/model-provider/llamacpp)
-- JS/TS (Programmable Prompt Engine CLI): [offline-ai/cli](https://github.com/offline-ai/cli)
-- JavaScript/Wasm (works in browser): [tangledgroup/llama-cpp-wasm](https://github.com/tangledgroup/llama-cpp-wasm)
-- Typescript/Wasm (nicer API, available on npm): [ngxson/wllama](https://github.com/ngxson/wllama)
-- Ruby: [yoshoku/llama_cpp.rb](https://github.com/yoshoku/llama_cpp.rb)
-- Ruby: [docusealco/rllama](https://github.com/docusealco/rllama)
-- Rust (more features): [edgenai/llama_cpp-rs](https://github.com/edgenai/llama_cpp-rs)
-- Rust (nicer API): [mdrokz/rust-llama.cpp](https://github.com/mdrokz/rust-llama.cpp)
-- Rust (more direct bindings): [utilityai/llama-cpp-rs](https://github.com/utilityai/llama-cpp-rs)
-- Rust (automated build from crates.io): [ShelbyJenkins/llm_client](https://github.com/ShelbyJenkins/llm_client)
-- C#/.NET: [SciSharp/LLamaSharp](https://github.com/SciSharp/LLamaSharp)
-- C#/VB.NET (more features - community license): [LM-Kit.NET](https://docs.lm-kit.com/lm-kit-net/index.html)
-- Scala 3: [donderom/llm4s](https://github.com/donderom/llm4s)
-- Clojure: [phronmophobic/llama.clj](https://github.com/phronmophobic/llama.clj)
-- React Native: [mybigday/llama.rn](https://github.com/mybigday/llama.rn)
-- Java: [kherud/java-llama.cpp](https://github.com/kherud/java-llama.cpp)
-- Java: [QuasarByte/llama-cpp-jna](https://github.com/QuasarByte/llama-cpp-jna)
-- Zig: [deins/llama.cpp.zig](https://github.com/Deins/llama.cpp.zig)
-- Flutter/Dart: [netdur/llama_cpp_dart](https://github.com/netdur/llama_cpp_dart)
-- Flutter: [xuegao-tzx/Fllama](https://github.com/xuegao-tzx/Fllama)
-- PHP (API bindings and features built on top of llama.cpp): [distantmagic/resonance](https://github.com/distantmagic/resonance) [(more info)](https://github.com/ggml-org/llama.cpp/pull/6326)
-- Guile Scheme: [guile_llama_cpp](https://savannah.nongnu.org/projects/guile-llama-cpp)
-- Swift [srgtuszy/llama-cpp-swift](https://github.com/srgtuszy/llama-cpp-swift)
-- Swift [ShenghaiWang/SwiftLlama](https://github.com/ShenghaiWang/SwiftLlama)
-- Delphi [Embarcadero/llama-cpp-delphi](https://github.com/Embarcadero/llama-cpp-delphi)
-- Go (no CGo needed): [hybridgroup/yzma](https://github.com/hybridgroup/yzma)
-- Android: [llama.android](/examples/llama.android)
-
-</details>
-
-<details>
-<summary>UIs</summary>
-
-*(to have a project listed here, it should clearly state that it depends on `llama.cpp`)*
-
-- [AI Sublime Text plugin](https://github.com/yaroslavyaroslav/OpenAI-sublime-text) (MIT)
-- [BonzAI App](https://apps.apple.com/us/app/bonzai-your-local-ai-agent/id6752847988) (proprietary)
-- [cztomsik/ava](https://github.com/cztomsik/ava) (MIT)
-- [Dot](https://github.com/alexpinel/Dot) (GPL)
-- [eva](https://github.com/ylsdamxssjxxdd/eva) (MIT)
-- [iohub/collama](https://github.com/iohub/coLLaMA) (Apache-2.0)
-- [janhq/jan](https://github.com/janhq/jan) (AGPL)
-- [johnbean393/Sidekick](https://github.com/johnbean393/Sidekick) (MIT)
-- [KanTV](https://github.com/zhouwg/kantv?tab=readme-ov-file) (Apache-2.0)
-- [KodiBot](https://github.com/firatkiral/kodibot) (GPL)
-- [llama.vim](https://github.com/ggml-org/llama.vim) (MIT)
-- [LARS](https://github.com/abgulati/LARS) (AGPL)
-- [Llama Assistant](https://github.com/vietanhdev/llama-assistant) (GPL)
-- [LlamaLib](https://github.com/undreamai/LlamaLib) (Apache-2.0)
-- [LLMFarm](https://github.com/guinmoon/LLMFarm?tab=readme-ov-file) (MIT)
-- [LLMUnity](https://github.com/undreamai/LLMUnity) (MIT)
-- [LMStudio](https://lmstudio.ai/) (proprietary)
-- [LocalAI](https://github.com/mudler/LocalAI) (MIT)
-- [LostRuins/koboldcpp](https://github.com/LostRuins/koboldcpp) (AGPL)
-- [MindMac](https://mindmac.app) (proprietary)
-- [MindWorkAI/AI-Studio](https://github.com/MindWorkAI/AI-Studio) (FSL-1.1-MIT)
-- [Mobile-Artificial-Intelligence/maid](https://github.com/Mobile-Artificial-Intelligence/maid) (MIT)
-- [Mozilla-Ocho/llamafile](https://github.com/Mozilla-Ocho/llamafile) (Apache-2.0)
-- [nat/openplayground](https://github.com/nat/openplayground) (MIT)
-- [nomic-ai/gpt4all](https://github.com/nomic-ai/gpt4all) (MIT)
-- [ollama/ollama](https://github.com/ollama/ollama) (MIT)
-- [oobabooga/text-generation-webui](https://github.com/oobabooga/text-generation-webui) (AGPL)
-- [PocketPal AI](https://github.com/a-ghorbani/pocketpal-ai) (MIT)
-- [psugihara/FreeChat](https://github.com/psugihara/FreeChat) (MIT)
-- [ptsochantaris/emeltal](https://github.com/ptsochantaris/emeltal) (MIT)
-- [pythops/tenere](https://github.com/pythops/tenere) (AGPL)
-- [ramalama](https://github.com/containers/ramalama) (MIT)
-- [semperai/amica](https://github.com/semperai/amica) (MIT)
-- [withcatai/catai](https://github.com/withcatai/catai) (MIT)
-- [Autopen](https://github.com/blackhole89/autopen) (GPL)
-
-</details>
-
-<details>
-<summary>Tools</summary>
-
-- [akx/ggify](https://github.com/akx/ggify) – download PyTorch models from Hugging Face Hub and convert them to GGML
-- [akx/ollama-dl](https://github.com/akx/ollama-dl) – download models from the Ollama library to be used directly with llama.cpp
-- [crashr/gppm](https://github.com/crashr/gppm) – launch llama.cpp instances utilizing NVIDIA Tesla P40 or P100 GPUs with reduced idle power consumption
-- [gpustack/gguf-parser](https://github.com/gpustack/gguf-parser-go/tree/main/cmd/gguf-parser) - review/check the GGUF file and estimate the memory usage
-- [Styled Lines](https://marketplace.unity.com/packages/tools/generative-ai/styled-lines-llama-cpp-model-292902) (proprietary licensed, async wrapper of inference part for game development in Unity3d with pre-built Mobile and Web platform wrappers and a model example)
-- [unslothai/unsloth](https://github.com/unslothai/unsloth) – 🦥 exports/saves fine-tuned and trained models to GGUF (Apache-2.0)
-
-</details>
-
-<details>
-<summary>Infrastructure</summary>
-
-- [Paddler](https://github.com/intentee/paddler) - Open-source LLMOps platform for hosting and scaling AI in your own infrastructure
-- [GPUStack](https://github.com/gpustack/gpustack) - Manage GPU clusters for running LLMs
-- [llama_cpp_canister](https://github.com/onicai/llama_cpp_canister) - llama.cpp as a smart contract on the Internet Computer, using WebAssembly
-- [llama-swap](https://github.com/mostlygeek/llama-swap) - transparent proxy that adds automatic model switching with llama-server
-- [Kalavai](https://github.com/kalavai-net/kalavai-client) - Crowdsource end to end LLM deployment at any scale
-- [llmaz](https://github.com/InftyAI/llmaz) - ☸️ Easy, advanced inference platform for large language models on Kubernetes.
-- [LLMKube](https://github.com/defilantech/llmkube) - Kubernetes operator for llama.cpp with multi-GPU and Apple Silicon Metal
-  support"
-</details>
-
-<details>
-<summary>Games</summary>
-
-- [Lucy's Labyrinth](https://github.com/MorganRO8/Lucys_Labyrinth) - A simple maze game where agents controlled by an AI model will try to trick you.
-
-</details>
-
+The `llama.cpp` project is build on top of the [ggml](https://github.com/ggml-org/ggml) library.
 
 ## Supported backends
 
 | Backend | Target devices |
 | --- | --- |
-| [Metal](docs/build.md#metal-build) | Apple Silicon |
 | [BLAS](docs/build.md#blas-build) | All |
 | [BLIS](docs/backend/BLIS.md) | All |
-| [SYCL](docs/backend/SYCL.md) | Intel GPU |
-| [OpenVINO [In Progress]](docs/backend/OPENVINO.md) | Intel CPUs, GPUs, and NPUs |
-| [MUSA](docs/build.md#musa) | Moore Threads GPU |
+| [CANN](docs/build.md#cann) | Ascend NPU |
 | [CUDA](docs/build.md#cuda) | Nvidia GPU |
 | [HIP](docs/build.md#hip) | AMD GPU |
-| [ZenDNN](docs/build.md#zendnn) | AMD CPU |
-| [Vulkan](docs/build.md#vulkan) | GPU |
-| [CANN](docs/build.md#cann) | Ascend NPU |
-| [OpenCL](docs/backend/OPENCL.md) | Adreno GPU |
-| [IBM zDNN](docs/backend/zDNN.md) | IBM Z & LinuxONE |
-| [WebGPU](docs/build.md#webgpu) | All |
-| [RPC](https://github.com/ggml-org/llama.cpp/tree/master/tools/rpc) | All |
 | [Hexagon [In Progress]](docs/backend/snapdragon/README.md) | Snapdragon |
+| [IBM zDNN](docs/backend/zDNN.md) | IBM Z & LinuxONE |
+| [MUSA](docs/build.md#musa) | Moore Threads GPU |
+| [Metal](docs/build.md#metal-build) | Apple Silicon |
+| [OpenCL](docs/backend/OPENCL.md) | Adreno GPU |
+| [OpenVINO [In Progress]](docs/backend/OPENVINO.md) | Intel CPUs, GPUs, and NPUs |
+| [RPC](https://github.com/ggml-org/llama.cpp/tree/master/tools/rpc) | All |
+| [SYCL](docs/backend/SYCL.md) | Intel GPU |
 | [VirtGPU](docs/backend/VirtGPU.md) | VirtGPU APIR |
+| [Vulkan](docs/build.md#vulkan) | GPU |
+| [WebGPU](docs/build.md#webgpu) | All |
+| [ZenDNN](docs/build.md#zendnn) | AMD CPU |
 
-## Obtaining and quantizing models
+## Documentation
 
-The [Hugging Face](https://huggingface.co) platform hosts a [number of LLMs](https://huggingface.co/models?library=gguf&sort=trending) compatible with `llama.cpp`:
-
-- [Trending](https://huggingface.co/models?library=gguf&sort=trending)
-- [LLaMA](https://huggingface.co/models?sort=trending&search=llama+gguf)
-
-You can either manually download the GGUF file or directly use any `llama.cpp`-compatible models from [Hugging Face](https://huggingface.co/) or other model hosting sites, by using this CLI argument: `-hf <user>/<model>[:quant]`. For example:
-
-```sh
-llama-cli -hf ggml-org/gemma-3-1b-it-GGUF
-```
-
-By default, the CLI would download from Hugging Face, you can switch to other options with the environment variable `MODEL_ENDPOINT`. The `MODEL_ENDPOINT` must point to a Hugging Face compatible API endpoint.
-
-After downloading a model, use the CLI tools to run it locally - see below.
-
-`llama.cpp` requires the model to be stored in the [GGUF](https://github.com/ggml-org/ggml/blob/master/docs/gguf.md) file format. Models in other data formats can be converted to GGUF using the `convert_*.py` Python scripts in this repo.
-
-The Hugging Face platform provides a variety of online tools for converting, quantizing and hosting models with `llama.cpp`:
-
-- Use the [GGUF-my-repo space](https://huggingface.co/spaces/ggml-org/gguf-my-repo) to convert to GGUF format and quantize model weights to smaller sizes
-- Use the [GGUF-my-LoRA space](https://huggingface.co/spaces/ggml-org/gguf-my-lora) to convert LoRA adapters to GGUF format (more info: https://github.com/ggml-org/llama.cpp/discussions/10123)
-- Use the [GGUF-editor space](https://huggingface.co/spaces/CISCai/gguf-editor) to edit GGUF meta data in the browser (more info: https://github.com/ggml-org/llama.cpp/discussions/9268)
-- Use the [Inference Endpoints](https://ui.endpoints.huggingface.co/) to directly host `llama.cpp` in the cloud (more info: https://github.com/ggml-org/llama.cpp/discussions/9669)
-
-To learn more about model quantization, [read this documentation](tools/quantize/README.md)
-
-## [`llama-cli`](tools/cli)
-
-#### A CLI tool for accessing and experimenting with most of `llama.cpp`'s functionality.
-
-- <details open>
-    <summary>Run in conversation mode</summary>
-
-    Models with a built-in chat template will automatically activate conversation mode. If this doesn't occur, you can manually enable it by adding `-cnv` and specifying a suitable chat template with `--chat-template NAME`
-
-    ```bash
-    llama-cli -m model.gguf
-
-    # > hi, who are you?
-    # Hi there! I'm your helpful assistant! I'm an AI-powered chatbot designed to assist and provide information to users like you. I'm here to help answer your questions, provide guidance, and offer support on a wide range of topics. I'm a friendly and knowledgeable AI, and I'm always happy to help with anything you need. What's on your mind, and how can I assist you today?
-    #
-    # > what is 1+1?
-    # Easy peasy! The answer to 1+1 is... 2!
-    ```
-
-    </details>
-
-- <details>
-    <summary>Run in conversation mode with custom chat template</summary>
-
-    ```bash
-    # use the "chatml" template (use -h to see the list of supported templates)
-    llama-cli -m model.gguf -cnv --chat-template chatml
-
-    # use a custom template
-    llama-cli -m model.gguf -cnv --in-prefix 'User: ' --reverse-prompt 'User:'
-    ```
-
-    </details>
-
-- <details>
-    <summary>Constrain the output with a custom grammar</summary>
-
-    ```bash
-    llama-cli -m model.gguf -n 256 --grammar-file grammars/json.gbnf -p 'Request: schedule a call at 8pm; Command:'
-
-    # {"appointmentTime": "8pm", "appointmentDetails": "schedule a a call"}
-    ```
-
-    The [grammars/](grammars/) folder contains a handful of sample grammars. To write your own, check out the [GBNF Guide](grammars/README.md).
-
-    For authoring more complex JSON grammars, check out https://grammar.intrinsiclabs.ai/
-
-    </details>
-
-
-## [`llama-server`](tools/server)
-
-#### A lightweight, [OpenAI API](https://github.com/openai/openai-openapi) compatible, HTTP server for serving LLMs.
-
-- <details open>
-    <summary>Start a local HTTP server with default configuration on port 8080</summary>
-
-    ```bash
-    llama-server -m model.gguf --port 8080
-
-    # Basic web UI can be accessed via browser: http://localhost:8080
-    # Chat completion endpoint: http://localhost:8080/v1/chat/completions
-    ```
-
-    </details>
-
-- <details>
-    <summary>Support multiple-users and parallel decoding</summary>
-
-    ```bash
-    # up to 4 concurrent requests, each with 4096 max context
-    llama-server -m model.gguf -c 16384 -np 4
-    ```
-
-    </details>
-
-- <details>
-    <summary>Enable speculative decoding</summary>
-
-    ```bash
-    # the draft.gguf model should be a small variant of the target model.gguf
-    llama-server -m model.gguf -md draft.gguf
-    ```
-
-    </details>
-
-- <details>
-    <summary>Serve an embedding model</summary>
-
-    ```bash
-    # use the /embedding endpoint
-    llama-server -m model.gguf --embedding --pooling cls -ub 8192
-    ```
-
-    </details>
-
-- <details>
-    <summary>Serve a reranking model</summary>
-
-    ```bash
-    # use the /reranking endpoint
-    llama-server -m model.gguf --reranking
-    ```
-
-    </details>
-
-- <details>
-    <summary>Constrain all outputs with a grammar</summary>
-
-    ```bash
-    # custom grammar
-    llama-server -m model.gguf --grammar-file grammar.gbnf
-
-    # JSON
-    llama-server -m model.gguf --grammar-file grammars/json.gbnf
-    ```
-
-    </details>
-
-
-## [`llama-perplexity`](tools/perplexity)
-
-#### A tool for measuring the [perplexity](tools/perplexity/README.md) [^1] (and other quality metrics) of a model over a given text.
-
-- <details open>
-    <summary>Measure the perplexity over a text file</summary>
-
-    ```bash
-    llama-perplexity -m model.gguf -f file.txt
-
-    # [1]15.2701,[2]5.4007,[3]5.3073,[4]6.2965,[5]5.8940,[6]5.6096,[7]5.7942,[8]4.9297, ...
-    # Final estimate: PPL = 5.4007 +/- 0.67339
-    ```
-
-    </details>
-
-- <details>
-    <summary>Measure KL divergence</summary>
-
-    ```bash
-    # TODO
-    ```
-
-    </details>
-
-[^1]: [https://huggingface.co/docs/transformers/perplexity](https://huggingface.co/docs/transformers/perplexity)
-
-## [`llama-bench`](tools/llama-bench)
-
-#### Benchmark the performance of the inference for various parameters.
-
-- <details open>
-    <summary>Run default benchmark</summary>
-
-    ```bash
-    llama-bench -m model.gguf
-
-    # Output:
-    # | model               |       size |     params | backend    | threads |          test |                  t/s |
-    # | ------------------- | ---------: | ---------: | ---------- | ------: | ------------: | -------------------: |
-    # | qwen2 1.5B Q4_0     | 885.97 MiB |     1.54 B | Metal,BLAS |      16 |         pp512 |      5765.41 ± 20.55 |
-    # | qwen2 1.5B Q4_0     | 885.97 MiB |     1.54 B | Metal,BLAS |      16 |         tg128 |        197.71 ± 0.81 |
-    #
-    # build: 3e0ba0e60 (4229)
-    ```
-
-    </details>
-
-## [`llama-simple`](examples/simple)
-
-#### A minimal example for implementing apps with `llama.cpp`. Useful for developers.
-
-- <details>
-    <summary>Basic text completion</summary>
-
-    ```bash
-    llama-simple -m model.gguf
-
-    # Hello my name is Kaitlyn and I am a 16 year old girl. I am a junior in high school and I am currently taking a class called "The Art of
-    ```
-
-    </details>
-
-
-## Contributing
-
-- Contributors can open PRs
-- Collaborators will be invited based on contributions
-- Maintainers can push to branches in the `llama.cpp` repo and merge PRs into the `master` branch
-- Any help with managing issues, PRs and projects is very appreciated!
-- See [good first issues](https://github.com/ggml-org/llama.cpp/issues?q=is%3Aissue+is%3Aopen+label%3A%22good+first+issue%22) for tasks suitable for first contributions
-- Read the [CONTRIBUTING.md](CONTRIBUTING.md) for more information
-- Make sure to read this: [Inference at the edge](https://github.com/ggml-org/llama.cpp/discussions/205)
-- A bit of backstory for those who are interested: [Changelog podcast](https://changelog.com/podcast/532)
-
-## Other documentation
+#### Tools
 
 - [cli](tools/cli/README.md)
 - [completion](tools/completion/README.md)
 - [server](tools/server/README.md)
 - [GBNF grammars](grammars/README.md)
 
-#### Development documentation
+#### Development
 
 - [How to build](docs/build.md)
 - [Running on Docker](docs/docker.md)
@@ -539,63 +103,19 @@ To learn more about model quantization, [read this documentation](tools/quantize
 - [Multi-GPU usage](docs/multi-gpu.md)
 - [Performance troubleshooting](docs/development/token_generation_performance_tips.md)
 - [GGML tips & tricks](https://github.com/ggml-org/llama.cpp/wiki/GGML-Tips-&-Tricks)
+- [XCFramework](docs/xcframework.md)
+- [Completions](docs/completions.md)
+- [Models](docs/models.md)
 
-#### Seminal papers and background on the models
+## Contributing
 
-If your issue is with model generation quality, then please at least scan the following links and papers to understand the limitations of LLaMA models. This is especially important when choosing an appropriate model size and appreciating both the significant and subtle differences between LLaMA models and ChatGPT:
-- LLaMA:
-    - [Introducing LLaMA: A foundational, 65-billion-parameter large language model](https://ai.facebook.com/blog/large-language-model-llama-meta-ai/)
-    - [LLaMA: Open and Efficient Foundation Language Models](https://arxiv.org/abs/2302.13971)
-- GPT-3
-    - [Language Models are Few-Shot Learners](https://arxiv.org/abs/2005.14165)
-- GPT-3.5 / InstructGPT / ChatGPT:
-    - [Aligning language models to follow instructions](https://openai.com/research/instruction-following)
-    - [Training language models to follow instructions with human feedback](https://arxiv.org/abs/2203.02155)
+- Contributors can open PRs
+- Collaborators will be invited based on contributions
+- Maintainers can push to branches in the `llama.cpp` repo and merge PRs into the `master` branch
+- Any help with managing issues, PRs and projects is very appreciated!
+- Read the [CONTRIBUTING.md](CONTRIBUTING.md) for more information
 
-## XCFramework
-The XCFramework is a precompiled version of the library for iOS, visionOS, tvOS,
-and macOS. It can be used in Swift projects without the need to compile the
-library from source. For example:
-```swift
-// swift-tools-version: 5.10
-// The swift-tools-version declares the minimum version of Swift required to build this package.
-
-import PackageDescription
-
-let package = Package(
-    name: "MyLlamaPackage",
-    targets: [
-        .executableTarget(
-            name: "MyLlamaPackage",
-            dependencies: [
-                "LlamaFramework"
-            ]),
-        .binaryTarget(
-            name: "LlamaFramework",
-            url: "https://github.com/ggml-org/llama.cpp/releases/download/b5046/llama-b5046-xcframework.zip",
-            checksum: "c19be78b5f00d8d29a25da41042cb7afa094cbf6280a225abe614b03b20029ab"
-        )
-    ]
-)
-```
-The above example is using an intermediate build `b5046` of the library. This can be modified
-to use a different version by changing the URL and checksum.
-
-## Completions
-Command-line completion is available for some environments.
-
-#### Bash Completion
-```bash
-$ build/bin/llama-cli --completion-bash > ~/.llama-completion.bash
-$ source ~/.llama-completion.bash
-```
-Optionally this can be added to your `.bashrc` or `.bash_profile` to load it
-automatically. For example:
-```console
-$ echo "source ~/.llama-completion.bash" >> ~/.bashrc
-```
-
-## Dependencies
+## Acknowledgements
 
 - [yhirose/cpp-httplib](https://github.com/yhirose/cpp-httplib) - Single-header HTTP server, used by `llama-server` - MIT license
 - [stb-image](https://github.com/nothings/stb) - Single-header image format decoder, used by multimodal subsystem - Public domain
diff --git a/docs/completions.md b/docs/completions.md
new file mode 100644
index 000000000..5376644c0
--- /dev/null
+++ b/docs/completions.md
@@ -0,0 +1,17 @@
+# Completions
+
+Command-line completion is available for some environments.
+
+## Bash Completion
+
+```bash
+$ build/bin/llama-cli --completion-bash > ~/.llama-completion.bash
+$ source ~/.llama-completion.bash
+```
+
+Optionally this can be added to your `.bashrc` or `.bash_profile` to load it
+automatically. For example:
+
+```console
+$ echo "source ~/.llama-completion.bash" >> ~/.bashrc
+```
diff --git a/docs/models.md b/docs/models.md
new file mode 100644
index 000000000..eee595263
--- /dev/null
+++ b/docs/models.md
@@ -0,0 +1,26 @@
+# Obtaining and quantizing models
+
+The [Hugging Face](https://huggingface.co) platform hosts [thousands of models](https://huggingface.co/models?library=gguf&sort=trending) compatible with `llama.cpp`:
+
+- [Trending](https://huggingface.co/models?library=gguf&sort=trending)
+
+You can use any `llama.cpp`-compatible model from [Hugging Face](https://huggingface.co/) using this CLI argument: `-hf <user>/<model>[:quant]`. For example:
+
+```sh
+llama cli -hf ggml-org/gemma-3-1b-it-GGUF
+```
+
+You can use the same CLI invocation to download from other sites, by pointing the `MODEL_ENDPOINT` environment variable to an endpoint compatible with the Hugging Face API.
+`llama.cpp` can also run models you have downloaded locally to your filesystem.
+
+After downloading a model, use the CLI tools to run it locally - see below.
+
+`llama.cpp` requires the model to be stored in the [GGUF](https://github.com/ggml-org/ggml/blob/master/docs/gguf.md) file format. Models in other data formats can be converted to GGUF using the `convert_*.py` Python scripts in this repo.
+To learn more about model quantization, [read this documentation](../tools/quantize/README.md)
+
+The Hugging Face platform provides a variety of online tools for converting, quantizing and hosting models with `llama.cpp`:
+
+- Use the [GGUF-my-repo space](https://huggingface.co/spaces/ggml-org/gguf-my-repo) to convert to GGUF format and quantize model weights to smaller sizes
+- Use the [GGUF-my-LoRA space](https://huggingface.co/spaces/ggml-org/gguf-my-lora) to convert LoRA adapters to GGUF format (more info: https://github.com/ggml-org/llama.cpp/discussions/10123)
+- Use the [GGUF-editor space](https://huggingface.co/spaces/CISCai/gguf-editor) to edit GGUF meta data in the browser (more info: https://github.com/ggml-org/llama.cpp/discussions/9268)
+- Use the [Inference Endpoints](https://ui.endpoints.huggingface.co/) to directly host `llama.cpp` in the cloud (more info: https://github.com/ggml-org/llama.cpp/discussions/9669)
diff --git a/docs/xcframework.md b/docs/xcframework.md
new file mode 100644
index 000000000..83c8b5004
--- /dev/null
+++ b/docs/xcframework.md
@@ -0,0 +1,31 @@
+# XCFramework
+
+The XCFramework is a precompiled version of the library for iOS, visionOS, tvOS,
+and macOS. It can be used in Swift projects without the need to compile the
+library from source. For example:
+
+```swift
+// swift-tools-version: 5.10
+// The swift-tools-version declares the minimum version of Swift required to build this package.
+
+import PackageDescription
+
+let package = Package(
+    name: "MyLlamaPackage",
+    targets: [
+        .executableTarget(
+            name: "MyLlamaPackage",
+            dependencies: [
+                "LlamaFramework"
+            ]),
+        .binaryTarget(
+            name: "LlamaFramework",
+            url: "https://github.com/ggml-org/llama.cpp/releases/download/b5046/llama-b5046-xcframework.zip",
+            checksum: "c19be78b5f00d8d29a25da41042cb7afa094cbf6280a225abe614b03b20029ab"
+        )
+    ]
+)
+```
+
+The above example is using an intermediate build `b5046` of the library. This can be modified
+to use a different version by changing the URL and checksum.

From e1a1abb78746c025f5e9039f590e37ccdb758ae7 Mon Sep 17 00:00:00 2001
From: Robert Esclapez <Robert.Garcia@amd.com>
Date: Thu, 30 Jul 2026 15:39:46 +0200
Subject: [PATCH 14/30] ggml-cuda: Allow transpose-free gemmv computation
 (#26171)

When matrix's weights are shaped 1xK is leverage a transpose-free
computation to use mat_mul_vec_f.
---
 ggml/src/ggml-cuda/ggml-cuda.cu | 14 ++++++++++++++
 tests/test-backend-ops.cpp      |  4 ++++
 2 files changed, 18 insertions(+)

diff --git a/ggml/src/ggml-cuda/ggml-cuda.cu b/ggml/src/ggml-cuda/ggml-cuda.cu
index d22b5657d..3e2c62295 100644
--- a/ggml/src/ggml-cuda/ggml-cuda.cu
+++ b/ggml/src/ggml-cuda/ggml-cuda.cu
@@ -1836,6 +1836,20 @@ static void ggml_cuda_mul_mat(ggml_backend_cuda_context & ctx, const ggml_tensor
         ggml_cuda_mul_mat_vec_f(ctx, src0, src1, nullptr, dst);
         return;
     }
+    // A transposed vector can still use MMVQ (i.e. ne01 == 1)
+    if (ne01 == 1 && ne11 > MMVF_MAX_BATCH_SIZE && ne2 == 1 && ne3 == 1
+            && src0->type == GGML_TYPE_F32
+            && ggml_is_contiguous(src0) && ggml_is_contiguous(src1) && ggml_is_contiguous(dst)
+            && ggml_cuda_should_use_mmvf(src1->type, cc, src1->ne, src1->nb, /*ne11 =*/ 1)) {
+        ggml_tensor dst_vec = *dst;
+        dst_vec.ne[0] = ne11;
+        dst_vec.ne[1] = 1;
+        dst_vec.nb[1] = dst_vec.nb[0]*ne11;
+        dst_vec.nb[2] = dst_vec.nb[1];
+        dst_vec.nb[3] = dst_vec.nb[1];
+        ggml_cuda_mul_mat_vec_f(ctx, src1, src0, nullptr, &dst_vec);
+        return;
+    }
     if (ggml_cuda_should_use_mmf(src0->type, cc, warp_size, src0->ne, src0->nb, ne11, /*mul_mat_id =*/ false)) {
         ggml_cuda_mul_mat_f(ctx, src0, src1, nullptr, dst);
         return;
diff --git a/tests/test-backend-ops.cpp b/tests/test-backend-ops.cpp
index ef18ecef8..e8d0e00d5 100644
--- a/tests/test-backend-ops.cpp
+++ b/tests/test-backend-ops.cpp
@@ -8838,6 +8838,10 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
     test_cases.emplace_back(new test_mul_mat(GGML_TYPE_Q8_0, GGML_TYPE_F32, 2880, 32, 2880, {1, 1}, {1, 1}));
     test_cases.emplace_back(new test_mul_mat(GGML_TYPE_MXFP4, GGML_TYPE_F32, 2880, 32, 2880, {1, 1}, {1, 1}));
 
+    // m == 1, with n on both sides of MMVF_MAX_BATCH_SIZE (8): mmvf below, operand swap above
+    for (int64_t n : {1, 7, 8, 9, 16, 128, 512}) {
+        test_cases.emplace_back(new test_mul_mat(GGML_TYPE_F32, GGML_TYPE_F32, 1, n, 2048, {1, 1}, {1, 1}));
+    }
 
 #if 0
     {

From 47f686f53f2a20fe82d87d79fa5c835714ee8f46 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 30 Jul 2026 19:34:04 +0300
Subject: [PATCH 15/30] tests : avoid building get-model.cpp many times
 (#26317)

* tests : remove get-model.cpp

* tests : fix quant type selection
---
 common/common.cpp                             | 14 +++++++++++
 common/common.h                               |  3 +++
 tests/CMakeLists.txt                          |  7 +++---
 tests/get-model.cpp                           | 21 ----------------
 tests/get-model.h                             |  2 --
 ...{qwen3.5-27b.schema => qwen3.6-27b.schema} |  0
 tests/test-autorelease.cpp                    |  6 ++---
 tests/test-backend-sampler.cpp                |  3 +--
 tests/test-model-load-cancel.cpp              |  4 ++--
 tests/test-quant-type-selection.cpp           | 24 +++++++++----------
 tests/test-rset-release.cpp                   |  8 +++----
 11 files changed, 42 insertions(+), 50 deletions(-)
 delete mode 100644 tests/get-model.cpp
 delete mode 100644 tests/get-model.h
 rename tests/snapshots/{qwen3.5-27b.schema => qwen3.6-27b.schema} (100%)

diff --git a/common/common.cpp b/common/common.cpp
index 9d5d2834a..1a0aeef4e 100644
--- a/common/common.cpp
+++ b/common/common.cpp
@@ -1476,6 +1476,20 @@ std::string common_get_model_endpoint() {
     return model_endpoint;
 }
 
+char * common_get_model_or_exit(int argc, char * argv[]) {
+    if (argc > 1) {
+        return argv[1];
+    }
+
+    char * path = getenv("LLAMACPP_TEST_MODELFILE");
+    if (!path || strlen(path) == 0) {
+        fprintf(stderr, "\033[33mWARNING: No model file provided. Skipping this test. Set LLAMACPP_TEST_MODELFILE=<gguf_model_path> to silence this warning and run this test.\n\033[0m");
+        exit(EXIT_SUCCESS);
+    }
+
+    return path;
+}
+
 common_context_seq_rm_type common_context_can_seq_rm(llama_context * ctx) {
     auto * mem = llama_get_memory(ctx);
     if (mem == nullptr) {
diff --git a/common/common.h b/common/common.h
index 14889193d..919c0ea10 100644
--- a/common/common.h
+++ b/common/common.h
@@ -930,6 +930,9 @@ void common_set_adapter_lora(struct llama_context * ctx, std::vector<common_adap
 // model endpoint from env
 std::string common_get_model_endpoint();
 
+// for testing purposes
+char * common_get_model_or_exit(int, char*[]);
+
 //
 // Context utils
 //
diff --git a/tests/CMakeLists.txt b/tests/CMakeLists.txt
index 805b74472..881e55c75 100644
--- a/tests/CMakeLists.txt
+++ b/tests/CMakeLists.txt
@@ -87,7 +87,7 @@ function(llama_build_and_test source)
     set(multiValueArgs ARGS)
     cmake_parse_arguments(LLAMA_TEST "${options}" "${oneValueArgs}" "${multiValueArgs}" ${ARGN})
 
-    set(TEST_SOURCES ${source} ${LLAMA_TEST_UNPARSED_ARGUMENTS} get-model.cpp)
+    set(TEST_SOURCES ${source} ${LLAMA_TEST_UNPARSED_ARGUMENTS})
 
     if (NOT DEFINED LLAMA_TEST_LABEL)
         set(LLAMA_TEST_LABEL "main")
@@ -148,7 +148,7 @@ if (LLAMA_LLGUIDANCE)
     llama_build_and_test(test-grammar-llguidance.cpp ARGS ${PROJECT_SOURCE_DIR}/models/ggml-vocab-llama-bpe.gguf)
 endif ()
 
-llama_build(test-recurrent-state-rollback.cpp get-model.cpp)
+llama_build(test-recurrent-state-rollback.cpp)
 
 if (NOT WIN32 OR NOT BUILD_SHARED_LIBS)
     # these tests are disabled on Windows because they use internal functions not exported with LLAMA_API (when building with shared libraries)
@@ -279,8 +279,9 @@ llama_build_and_test(test-save-load-state.cpp LABEL "model" ARGS -m "${MODEL_DES
 set_tests_properties(test-save-load-state PROPERTIES FIXTURES_REQUIRED test-download-model)
 
 if (APPLE)
-    llama_build(test-rset-release.cpp get-model.cpp)
+    llama_build(test-rset-release.cpp)
 endif()
+
 if (NOT GGML_BACKEND_DL)
     # these tests use the backends directly and cannot be built with dynamic loading
     llama_build_and_test(test-barrier.cpp)
diff --git a/tests/get-model.cpp b/tests/get-model.cpp
deleted file mode 100644
index 4edb685f0..000000000
--- a/tests/get-model.cpp
+++ /dev/null
@@ -1,21 +0,0 @@
-#include <cstdio>
-#include <cstdlib>
-#include <cstring>
-
-#include "get-model.h"
-
-char * get_model_or_exit(int argc, char *argv[]) {
-    char * model_path;
-    if (argc > 1) {
-        model_path = argv[1];
-
-    } else {
-        model_path = getenv("LLAMACPP_TEST_MODELFILE");
-        if (!model_path || strlen(model_path) == 0) {
-            fprintf(stderr, "\033[33mWARNING: No model file provided. Skipping this test. Set LLAMACPP_TEST_MODELFILE=<gguf_model_path> to silence this warning and run this test.\n\033[0m");
-            exit(EXIT_SUCCESS);
-        }
-    }
-
-    return model_path;
-}
diff --git a/tests/get-model.h b/tests/get-model.h
deleted file mode 100644
index 81a3a0fef..000000000
--- a/tests/get-model.h
+++ /dev/null
@@ -1,2 +0,0 @@
-#pragma once
-char * get_model_or_exit(int, char*[]);
diff --git a/tests/snapshots/qwen3.5-27b.schema b/tests/snapshots/qwen3.6-27b.schema
similarity index 100%
rename from tests/snapshots/qwen3.5-27b.schema
rename to tests/snapshots/qwen3.6-27b.schema
diff --git a/tests/test-autorelease.cpp b/tests/test-autorelease.cpp
index ca87c56a8..370428809 100644
--- a/tests/test-autorelease.cpp
+++ b/tests/test-autorelease.cpp
@@ -1,15 +1,13 @@
 // ref: https://github.com/ggml-org/llama.cpp/issues/4952#issuecomment-1892864763
 
-#include <cstdio>
-#include <string>
 #include <thread>
 
 #include "llama.h"
-#include "get-model.h"
+#include "common.h"
 
 // This creates a new context inside a pthread and then tries to exit cleanly.
 int main(int argc, char ** argv) {
-    auto * model_path = get_model_or_exit(argc, argv);
+    auto * model_path = common_get_model_or_exit(argc, argv);
 
     std::thread([&model_path]() {
         llama_backend_init();
diff --git a/tests/test-backend-sampler.cpp b/tests/test-backend-sampler.cpp
index 61ddf91fe..c24076e31 100644
--- a/tests/test-backend-sampler.cpp
+++ b/tests/test-backend-sampler.cpp
@@ -1,7 +1,6 @@
 #include "ggml.h"
 #include "llama.h"
 #include "llama-cpp.h"
-#include "get-model.h"
 #include "common.h"
 
 #ifdef NDEBUG
@@ -1136,7 +1135,7 @@ int main(int argc, char ** argv) {
     test_args args = parse_cli(argc, argv);
 
     if (args.model.empty()) {
-        args.model = get_model_or_exit(1, argv);
+        args.model = common_get_model_or_exit(1, argv);
     }
 
     {
diff --git a/tests/test-model-load-cancel.cpp b/tests/test-model-load-cancel.cpp
index ecc302271..f8139b26d 100644
--- a/tests/test-model-load-cancel.cpp
+++ b/tests/test-model-load-cancel.cpp
@@ -1,10 +1,10 @@
 #include "llama.h"
-#include "get-model.h"
+#include "common.h"
 
 #include <cstdlib>
 
 int main(int argc, char *argv[] ) {
-    auto * model_path = get_model_or_exit(argc, argv);
+    auto * model_path = common_get_model_or_exit(argc, argv);
     auto * file = fopen(model_path, "r");
     if (file == nullptr) {
         fprintf(stderr, "no model at '%s' found\n", model_path);
diff --git a/tests/test-quant-type-selection.cpp b/tests/test-quant-type-selection.cpp
index 3c8983360..9a5f5e53e 100644
--- a/tests/test-quant-type-selection.cpp
+++ b/tests/test-quant-type-selection.cpp
@@ -216,18 +216,18 @@ static std::string snapshot_file_from_name(const std::string & name) {
 }
 
 static const remote_model_spec model_specs[] = {
-    { "ggml-org/Qwen3-0.6B-GGUF",                   "Q8_0"   },
-    { "ggml-org/GLM-4.6V-GGUF",                     "Q8_0"   },
-    { "ggml-org/Step-3.5-Flash-GGUF",               "Q4_K"   },
-    { "ggml-org/Qwen3-Coder-Next-GGUF",             "Q8_0"   },
-    { "ggml-org/Qwen3-14B-GGUF",                    "Q8_0"   },
-    { "ggml-org/Nemotron-Nano-3-30B-A3B-GGUF",      "Q8_0"   },
-    { "ggml-org/gpt-oss-120b-GGUF",                 "mxfp4"  },
-    { "ggml-org/gemma-3-4b-it-GGUF",                "Q8_0"   },
-    { "bartowski/Meta-Llama-3.1-70B-Instruct-GGUF", "Q4_K_M" },
-    { "bartowski/deepseek-ai_DeepSeek-V3.1-GGUF",   "IQ1_M"  },
-    { "bartowski/Qwen_Qwen3.5-397B-A17B-GGUF",      "IQ1_S"  }, // TODO: swap with ggml-org if/when it's released
-    { "bartowski/Qwen_Qwen3.5-27B-GGUF",            "Q8_0"   }, // TODO: swap with ggml-org if/when it's released
+    { "ggml-org/Qwen3-0.6B-GGUF",                     "Q8_0"   },
+    { "ggml-org/GLM-4.6V-GGUF",                       "Q8_0"   },
+    { "ggml-org/Step-3.5-Flash-GGUF",                 "Q4_K"   },
+    { "ggml-org/Qwen3-Coder-Next-GGUF",               "Q8_0"   },
+    { "ggml-org/Qwen3-14B-GGUF",                      "Q8_0"   },
+    { "ggml-org/NVIDIA-Nemotron-Nano-3-30B-A3B-GGUF", "Q8_0"   },
+    { "ggml-org/gpt-oss-120b-GGUF",                   "mxfp4"  },
+    { "ggml-org/gemma-3-4b-it-GGUF",                  "Q8_0"   },
+    { "bartowski/Meta-Llama-3.1-70B-Instruct-GGUF",   "Q4_K_M" },
+    { "bartowski/deepseek-ai_DeepSeek-V3.1-GGUF",     "IQ1_M"  },
+  //{ "bartowski/Qwen_Qwen3.5-397B-A17B-GGUF",        "IQ1_S"  }, // TODO: swap with ggml-org if/when it's released
+    { "ggml-org/Qwen3.6-27B-GGUF",                    "Q8_0"   },
 };
 
 static const int n_model_specs = (int) (sizeof(model_specs) / sizeof(model_specs[0]));
diff --git a/tests/test-rset-release.cpp b/tests/test-rset-release.cpp
index bf03c5e8b..c60801c11 100644
--- a/tests/test-rset-release.cpp
+++ b/tests/test-rset-release.cpp
@@ -3,14 +3,14 @@
 // thus, this test is not run by default
 // example model to run with: google/gemma-4-E4B-it-qat-q4_0-gguf
 
+#include "llama.h"
+#include "common.h"
+
 #include <cstdint>
 #include <mach/mach.h>
 #include <mach/mach_host.h>
 #include <unistd.h>
 
-#include "llama.h"
-#include "get-model.h"
-
 static uint64_t wired_memory() {
     vm_statistics64_data_t vmstat;
     mach_msg_type_number_t count = HOST_VM_INFO64_COUNT;
@@ -21,7 +21,7 @@ static uint64_t wired_memory() {
 }
 
 int main(int argc, char ** argv) {
-    auto * model_path = get_model_or_exit(argc, argv);
+    auto * model_path = common_get_model_or_exit(argc, argv);
 
     llama_backend_init();
 

From 432d7ffe2c3b4e539f3d0d4ae0a4893090a018d6 Mon Sep 17 00:00:00 2001
From: o7si <32285332+o7si@users.noreply.github.com>
Date: Fri, 31 Jul 2026 00:48:00 +0800
Subject: [PATCH 16/30] llama-context : sync pending async copies before
 clearing embd_seq (#25676)

---
 src/llama-context.cpp | 22 +++++++++++++++++-----
 1 file changed, 17 insertions(+), 5 deletions(-)

diff --git a/src/llama-context.cpp b/src/llama-context.cpp
index 9b399d609..3e0af3f31 100644
--- a/src/llama-context.cpp
+++ b/src/llama-context.cpp
@@ -474,6 +474,9 @@ llama_context::llama_context(
 }
 
 llama_context::~llama_context() {
+    // wait for any pending asynchronous copies into the output buffers before they are freed
+    synchronize();
+
     if (!model.hparams.no_alloc) {
         for (size_t i = 0; i < backend_ptrs.size(); ++i) {
             ggml_backend_t             backend = backend_ptrs[i];
@@ -1417,13 +1420,17 @@ int llama_context::encode(const llama_batch & batch_inp) {
     // micro-batching is not possible for non-causal encoding, so we process the batch in a single shot
     GGML_ASSERT(cparams.n_ubatch >= n_tokens && "encoder requires n_ubatch >= n_tokens");
 
+    // TODO: this clear of the buffer can easily be forgotten - need something better
+    // sync first so any in-flight async copies into embd_seq complete before it is freed
+    if (!embd_seq.empty()) {
+        synchronize();
+    }
+    embd_seq.clear();
+
     if (t_compute_start_us == 0) {
         t_compute_start_us = ggml_time_us();
     }
 
-    // TODO: this clear of the buffer can easily be forgotten - need something better
-    embd_seq.clear();
-
     sched_reserve();
 
     n_queued_tokens += n_tokens;
@@ -1762,13 +1769,18 @@ int llama_context::decode(const llama_batch & batch_inp) {
 
     GGML_ASSERT((cparams.causal_attn || cparams.n_ubatch >= n_tokens_all) && "non-causal attention requires n_ubatch >= n_tokens");
 
+    // TODO: this clear of the buffer can easily be forgotten - need something better
+    // sync first so any in-flight async copies into embd_seq complete before it is freed
+    if (!embd_seq.empty()) {
+        synchronize();
+    }
+    embd_seq.clear();
+
     if (t_compute_start_us == 0) {
         t_compute_start_us = ggml_time_us();
     }
     n_queued_tokens += n_tokens_all;
 
-    // TODO: this clear of the buffer can easily be forgotten - need something better
-    embd_seq.clear();
     output_swaps.clear();
 
     sched_reserve();

From 958d9c0b61d0025ca8e2fff32d66a1576173fde9 Mon Sep 17 00:00:00 2001
From: pmaybank <113125070+pmaybank@users.noreply.github.com>
Date: Thu, 30 Jul 2026 18:14:16 +0100
Subject: [PATCH 17/30] Test support for alternative conv layout (#25617)

* add  bool cwhn = true to conv_2d test cases

* add layout check at graph building time

* extend layout checks for conv2d.cu kernel

* in CPU back-end kernel needs to be stored contiguously to prevent test failures with cwhn=1

* trim white space

* do op support check in vulkan backend

* fix CI failure and vulkan run-time assert failure by introducing new graph build-time check in ggml_backend_vk_device_supports_op

* add additional check in support_op function for Vulkan to fix run-time assert failure
---
 ggml/src/ggml-cpu/ggml-cpu.cpp       |  2 ++
 ggml/src/ggml-cuda/conv2d.cu         |  1 +
 ggml/src/ggml-cuda/ggml-cuda.cu      |  2 +-
 ggml/src/ggml-vulkan/ggml-vulkan.cpp |  7 +++++++
 tests/test-backend-ops.cpp           | 19 +++++++++++++++----
 5 files changed, 26 insertions(+), 5 deletions(-)

diff --git a/ggml/src/ggml-cpu/ggml-cpu.cpp b/ggml/src/ggml-cpu/ggml-cpu.cpp
index 74631c285..16cc5116c 100644
--- a/ggml/src/ggml-cpu/ggml-cpu.cpp
+++ b/ggml/src/ggml-cpu/ggml-cpu.cpp
@@ -469,6 +469,8 @@ static bool ggml_backend_cpu_device_supports_op(ggml_backend_dev_t dev, const st
             return (src0->type == GGML_TYPE_F32 ||
                     ((src0->type == GGML_TYPE_F16 || ggml_is_quantized(src0->type)) && src0->ne[2] == src1->ne[2] && src0->ne[3] == src1->ne[3])) &&
                 src1->type == GGML_TYPE_F32 && op->type == GGML_TYPE_F32;
+        case GGML_OP_CONV_2D:
+            return ggml_is_contiguous(op->src[0]);
         default:
             return true;
     }
diff --git a/ggml/src/ggml-cuda/conv2d.cu b/ggml/src/ggml-cuda/conv2d.cu
index 142dd6690..14774d4a5 100644
--- a/ggml/src/ggml-cuda/conv2d.cu
+++ b/ggml/src/ggml-cuda/conv2d.cu
@@ -126,6 +126,7 @@ void ggml_cuda_op_conv2d(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
     const float *       X_D    = (const float *) input->data;
     float *             Y_D    = (float *) dst->data;
 
+    GGML_ASSERT(ggml_is_contiguous(input));
     GGML_ASSERT(ggml_is_contiguous(kernel));
     GGML_ASSERT(kernel->type == GGML_TYPE_F16 || kernel->type == GGML_TYPE_F32);
 
diff --git a/ggml/src/ggml-cuda/ggml-cuda.cu b/ggml/src/ggml-cuda/ggml-cuda.cu
index 3e2c62295..561ab7ac5 100644
--- a/ggml/src/ggml-cuda/ggml-cuda.cu
+++ b/ggml/src/ggml-cuda/ggml-cuda.cu
@@ -5105,7 +5105,7 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
         case GGML_OP_IM2COL:
         case GGML_OP_IM2COL_3D:
         case GGML_OP_CONV_2D:
-            return true;
+            return (ggml_is_contiguous(op->src[0]) && ggml_is_contiguous(op->src[1]));
         case GGML_OP_CONV_2D_DW:
             return op->src[0]->type == GGML_TYPE_F32;
         case GGML_OP_CONV_TRANSPOSE_2D:
diff --git a/ggml/src/ggml-vulkan/ggml-vulkan.cpp b/ggml/src/ggml-vulkan/ggml-vulkan.cpp
index 49fd6e0b4..8a1bff65a 100644
--- a/ggml/src/ggml-vulkan/ggml-vulkan.cpp
+++ b/ggml/src/ggml-vulkan/ggml-vulkan.cpp
@@ -18016,10 +18016,17 @@ static bool ggml_backend_vk_device_supports_op(ggml_backend_dev_t dev, const ggm
         case GGML_OP_CONV_2D:
         case GGML_OP_CONV_TRANSPOSE_2D:
             {
+                const bool transpose = op->op == GGML_OP_CONV_TRANSPOSE_2D;
+                const int64_t cout = !transpose ? op->src[0]->ne[3] : op->src[0]->ne[2];
+                const int64_t cin  = !transpose ? op->src[0]->ne[2] : op->src[0]->ne[3];
+
                 // Channel-contiguous format is not supported yet.
                 return ((op->src[0]->type == GGML_TYPE_F32 || op->src[0]->type == GGML_TYPE_F16) &&
+                    (op->src[0]->nb[0] == sizeof(float) || op->src[0]->nb[0] == sizeof(ggml_fp16_t) ) &&
                     op->src[1]->type == GGML_TYPE_F32 &&
                     op->type == GGML_TYPE_F32 &&
+                    cout == op->ne[2] &&
+                    cin == op->src[1]->ne[2] &&
                     ggml_is_contiguous(op->src[0]) &&
                     ggml_is_contiguous(op->src[1]) &&
                     ggml_is_contiguous(op));
diff --git a/tests/test-backend-ops.cpp b/tests/test-backend-ops.cpp
index e8d0e00d5..162bfc98c 100644
--- a/tests/test-backend-ops.cpp
+++ b/tests/test-backend-ops.cpp
@@ -8329,7 +8329,11 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
             test_cases.emplace_back(new test_conv_2d(
                 { act_case[iwh_idx], act_case[iwh_idx], act_case[Cin_idx], act_case[B_idx] },
                 { act_case[kwh_idx], act_case[kwh_idx], act_case[Cin_idx], act_case[Cout_idx] },
-                kernel_type, 1, 1, 0, 0, 1, 1, false));
+                kernel_type, 1, 1, 0, 0, 1, 1, false));  // bool cwhn = false
+            test_cases.emplace_back(new test_conv_2d(
+                { act_case[iwh_idx], act_case[iwh_idx], act_case[Cin_idx], act_case[B_idx] },
+                { act_case[kwh_idx], act_case[kwh_idx], act_case[Cin_idx], act_case[Cout_idx] },
+                kernel_type, 1, 1, 0, 0, 1, 1, true));  // bool cwhn = true
         }
     }
 #endif
@@ -8358,7 +8362,9 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
                                         calc_conv_output_size(H, KH, s1, p1, d1) > 0) {
                                         for (auto kernel_type : {GGML_TYPE_F32, GGML_TYPE_F16}) {
                                             test_cases.emplace_back(new test_conv_2d(
-                                                { W, H, Cin, 2 }, { KW, KH, Cin, Cout }, kernel_type, s0, s1, p0, p1, d0, d1, false));
+                                                { W, H, Cin, 2 }, { KW, KH, Cin, Cout }, kernel_type, s0, s1, p0, p1, d0, d1, false)); // bool cwhn = false
+                                            test_cases.emplace_back(new test_conv_2d(
+                                                { W, H, Cin, 2 }, { KW, KH, Cin, Cout }, kernel_type, s0, s1, p0, p1, d0, d1, true));  // bool cwhn = true
                                         }
                                     }
                                 }
@@ -8370,7 +8376,8 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
         }
     }
     for (auto kernel_type : {GGML_TYPE_F32, GGML_TYPE_F16}) {
-        test_cases.emplace_back(new test_conv_2d({ 256, 256, 192, 1 }, { 3, 3, 192, 96 }, kernel_type, 1, 1, 1, 1, 1, 1, false));
+        test_cases.emplace_back(new test_conv_2d({ 256, 256, 192, 1 }, { 3, 3, 192, 96 }, kernel_type, 1, 1, 1, 1, 1, 1, false)); // bool cwhn = false
+        test_cases.emplace_back(new test_conv_2d({ 256, 256, 192, 1 }, { 3, 3, 192, 96 }, kernel_type, 1, 1, 1, 1, 1, 1, true));  // bool cwhn = true
     }
 
     // sycl backend will limit task global_range < MAX_INT
@@ -9757,7 +9764,11 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_perf() {
             test_cases.emplace_back(new test_conv_2d(
                 { act_case[iwh_idx], act_case[iwh_idx], act_case[Cin_idx], act_case[B_idx] },
                 { act_case[kwh_idx], act_case[kwh_idx], act_case[Cin_idx], act_case[Cout_idx] },
-                kernel_type, 1, 1, 0, 0, 1, 1, false));
+                kernel_type, 1, 1, 0, 0, 1, 1, false));   // bool cwhn = false
+            test_cases.emplace_back(new test_conv_2d(
+                { act_case[iwh_idx], act_case[iwh_idx], act_case[Cin_idx], act_case[B_idx] },
+                { act_case[kwh_idx], act_case[kwh_idx], act_case[Cin_idx], act_case[Cout_idx] },
+                kernel_type, 1, 1, 0, 0, 1, 1, true));    // bool cwhn = true
         }
     }
 

From ea63b4d32ea1b66bdbe369be7f9443f6c00f8b31 Mon Sep 17 00:00:00 2001
From: Jeff Bolz <jbolz@nvidia.com>
Date: Thu, 30 Jul 2026 19:11:32 +0100
Subject: [PATCH 18/30] vulkan: Support quantized concat (#25684)

---
 ggml/src/ggml-vulkan/ggml-vulkan.cpp | 99 +++++++++++++++++++++-------
 1 file changed, 77 insertions(+), 22 deletions(-)

diff --git a/ggml/src/ggml-vulkan/ggml-vulkan.cpp b/ggml/src/ggml-vulkan/ggml-vulkan.cpp
index 8a1bff65a..e32dd7a56 100644
--- a/ggml/src/ggml-vulkan/ggml-vulkan.cpp
+++ b/ggml/src/ggml-vulkan/ggml-vulkan.cpp
@@ -1481,6 +1481,11 @@ struct vk_op_binary_push_constants {
     float param1; float param2; int32_t param3;
 };
 
+// Distinct type with the same layout so concat can overload tensor offset initialization.
+struct vk_op_concat_push_constants : vk_op_binary_push_constants {};
+static_assert(sizeof(vk_op_concat_push_constants) == sizeof(vk_op_binary_push_constants));
+static_assert(std::is_standard_layout_v<vk_op_concat_push_constants>);
+
 struct vk_op_multi_add_push_constants {
     // shape for dst
     uint32_t ne20; uint32_t ne21; uint32_t ne22; uint32_t ne23;
@@ -2246,6 +2251,40 @@ static uint32_t get_misalign_bytes(const ggml_backend_vk_context * ctx, const gg
     return ((vk_tensor_offset(t) + t->view_offs) & (ctx->device->properties.limits.minStorageBufferOffsetAlignment - 1));;
 }
 
+static uint32_t ggml_vk_concat_unit_size(ggml_type type) {
+    const uint32_t type_size = ggml_type_size(type);
+
+    if (!ggml_is_quantized(type)) {
+        return type_size;
+    }
+
+    // Use the widest existing concat shader that evenly divides a quant block.
+    if (type_size % 8 == 0) {
+        return 8;
+    }
+    if (type_size % 4 == 0) {
+        return 4;
+    }
+    if (type_size % 2 == 0) {
+        return 2;
+    }
+    return 1;
+}
+
+static bool ggml_vk_concat_supported(const ggml_tensor * src0, const ggml_tensor * src1, const ggml_tensor * dst) {
+    if (src0->type != src1->type || src0->type != dst->type) {
+        return false;
+    }
+
+    if (!ggml_is_quantized(src0->type)) {
+        const size_t type_size = ggml_type_size(src0->type);
+        return type_size == 1 || type_size == 2 || type_size == 4 || type_size == 8;
+    }
+
+    // Quantized tensor rows are block-aligned when created.
+    return ggml_is_contiguous_rows(src0) && ggml_is_contiguous_rows(src1) && ggml_is_contiguous_rows(dst);
+}
+
 template <typename T> void init_pushconst_tensor_offsets(ggml_backend_vk_context * ctx, T &p, const ggml_tensor * src0, const ggml_tensor * src1, const ggml_tensor * src2, const ggml_tensor * src3, ggml_tensor * dst) {
     GGML_UNUSED(p);
     GGML_UNUSED(src0);
@@ -10896,14 +10935,10 @@ static vk_pipeline ggml_vk_op_get_pipeline(ggml_backend_vk_context * ctx, const
         }
         return nullptr;
     case GGML_OP_CONCAT: {
-        if (src0->type != src1->type || src0->type != dst->type) {
+        if (!ggml_vk_concat_supported(src0, src1, dst)) {
             return nullptr;
         }
-        if (ggml_blck_size(src0->type) != 1) {
-            return nullptr;
-        }
-        const size_t type_size = ggml_type_size(src0->type);
-        switch (type_size) {
+        switch (ggml_vk_concat_unit_size(src0->type)) {
         case 1:
             return ctx->device->pipeline_concat_i8;
         case 2:
@@ -11595,6 +11630,18 @@ template <> void init_pushconst_tensor_offsets(ggml_backend_vk_context * ctx, vk
     GGML_UNUSED(src3);
 }
 
+template <> void init_pushconst_tensor_offsets(ggml_backend_vk_context * ctx, vk_op_concat_push_constants &p, const ggml_tensor * src0, const ggml_tensor * src1, const ggml_tensor * src2, const ggml_tensor * src3, ggml_tensor * dst) {
+    const uint32_t unit_size = ggml_vk_concat_unit_size(dst->type);
+    const uint32_t a_offset = get_misalign_bytes(ctx, src0) / unit_size;
+    const uint32_t b_offset = get_misalign_bytes(ctx, src1) / unit_size;
+    const uint32_t d_offset = get_misalign_bytes(ctx, dst) / unit_size;
+
+    p.misalign_offsets = (a_offset << 16) | (b_offset << 8) | d_offset;
+
+    GGML_UNUSED(src2);
+    GGML_UNUSED(src3);
+}
+
 template <> void init_pushconst_tensor_offsets(ggml_backend_vk_context * ctx, vk_op_upscale_push_constants &p, const ggml_tensor * src0, const ggml_tensor * src1, const ggml_tensor * src2, const ggml_tensor * src3, ggml_tensor * dst) {
     const uint32_t a_offset = get_misalign_bytes(ctx, src0) / ggml_type_size(src0->type);
     const uint32_t d_offset = get_misalign_bytes(ctx, dst) / ggml_type_size(dst->type);
@@ -11630,7 +11677,7 @@ static void ggml_vk_op_f32(ggml_backend_vk_context * ctx, vk_context& subctx, co
     }
     std::cerr << "), (" << dst << ", name=" << dst->name << ", type=" << dst->type << ", ne0=" << dst->ne[0] << ", ne1=" << dst->ne[1] << ", ne2=" << dst->ne[2] << ", ne3=" << dst->ne[3] << ", nb0=" << dst->nb[0] << ", nb1=" << dst->nb[1] << ", nb2=" << dst->nb[2] << ", nb3=" << dst->nb[3];
     std::cerr << "), " << ggml_op_name(op) << ")");
-    GGML_ASSERT(op == GGML_OP_GET_ROWS || op == GGML_OP_CPY || (!ggml_is_quantized(src0->type) && (src1 == nullptr || !ggml_is_quantized(src1->type))));  // NOLINT
+    GGML_ASSERT(op == GGML_OP_GET_ROWS || op == GGML_OP_CPY || op == GGML_OP_CONCAT || (!ggml_is_quantized(src0->type) && (src1 == nullptr || !ggml_is_quantized(src1->type))));  // NOLINT
     GGML_ASSERT(dst->buffer != nullptr);
     const uint64_t ne00 = src0->ne[0];
     const uint64_t ne01 = src0->ne[1];
@@ -11885,6 +11932,9 @@ static void ggml_vk_op_f32(ggml_backend_vk_context * ctx, vk_context& subctx, co
                     ne *= ggml_type_size(src0->type) / 2;
                 }
             }
+            if (op == GGML_OP_CONCAT && ggml_is_quantized(dst->type)) {
+                ne = ne / ggml_blck_size(dst->type) * ggml_type_size(dst->type) / ggml_vk_concat_unit_size(dst->type);
+            }
             // copy_to_quant has block size of 32, and each thread does QUANT_K elements.
             // Splitting into 512x512xZ wouldn't work well since each workgroup does 1024 elements.
             // So divide by block size here before splitting into 512x512 groups.
@@ -12525,18 +12575,28 @@ static void ggml_vk_opt_step_sgd(ggml_backend_vk_context * ctx, vk_context& subc
 static void ggml_vk_concat(ggml_backend_vk_context * ctx, vk_context& subctx, const ggml_tensor * src0, const ggml_tensor * src1, ggml_tensor * dst) {
     int * op_params = (int *)dst->op_params;
 
-    const uint32_t src0_type_size = ggml_type_size(src0->type);
-    const uint32_t src1_type_size = ggml_type_size(src1->type);
-    const uint32_t dst_type_size = ggml_type_size(dst->type);
+    const uint32_t unit_size = ggml_vk_concat_unit_size(dst->type);
+    const uint32_t units_per_block = ggml_type_size(dst->type) / unit_size;
+    const uint32_t block_size = ggml_blck_size(dst->type);
+    const bool quantized = ggml_is_quantized(dst->type);
 
-    ggml_vk_op_f32<vk_op_binary_push_constants>(ctx, subctx, src0, src1, nullptr, nullptr, dst, GGML_OP_CONCAT, {
-        (uint32_t)ggml_nelements(dst),
-        (uint32_t)src0->ne[0], (uint32_t)src0->ne[1], (uint32_t)src0->ne[2],(uint32_t)src0->ne[3], (uint32_t)src0->nb[0] / src0_type_size, (uint32_t)src0->nb[1] / src0_type_size, (uint32_t)src0->nb[2] / src0_type_size, (uint32_t)src0->nb[3] / src0_type_size,
-        (uint32_t)src1->ne[0], (uint32_t)src1->ne[1], (uint32_t)src1->ne[2],(uint32_t)src1->ne[3], (uint32_t)src1->nb[0] / src1_type_size, (uint32_t)src1->nb[1] / src1_type_size, (uint32_t)src1->nb[2] / src1_type_size, (uint32_t)src1->nb[3] / src1_type_size,
-        (uint32_t) dst->ne[0], (uint32_t) dst->ne[1], (uint32_t) dst->ne[2],(uint32_t) dst->ne[3], (uint32_t) dst->nb[0] /  dst_type_size, (uint32_t) dst->nb[1] /  dst_type_size, (uint32_t) dst->nb[2] /  dst_type_size, (uint32_t) dst->nb[3] /  dst_type_size,
+    // Address dimension 0 in packed storage units; higher strides may be noncontiguous.
+    const uint32_t ne00 = src0->ne[0] / block_size * units_per_block;
+    const uint32_t ne10 = src1->ne[0] / block_size * units_per_block;
+    const uint32_t ne20 =  dst->ne[0] / block_size * units_per_block;
+    const uint32_t nb00 = quantized ? 1 : src0->nb[0] / unit_size;
+    const uint32_t nb10 = quantized ? 1 : src1->nb[0] / unit_size;
+    const uint32_t nb20 = quantized ? 1 :  dst->nb[0] / unit_size;
+
+    vk_op_concat_push_constants pc {{
+        ne20 * (uint32_t)dst->ne[1] * (uint32_t)dst->ne[2] * (uint32_t)dst->ne[3],
+        ne00, (uint32_t)src0->ne[1], (uint32_t)src0->ne[2],(uint32_t)src0->ne[3], nb00, (uint32_t)src0->nb[1] / unit_size, (uint32_t)src0->nb[2] / unit_size, (uint32_t)src0->nb[3] / unit_size,
+        ne10, (uint32_t)src1->ne[1], (uint32_t)src1->ne[2],(uint32_t)src1->ne[3], nb10, (uint32_t)src1->nb[1] / unit_size, (uint32_t)src1->nb[2] / unit_size, (uint32_t)src1->nb[3] / unit_size,
+        ne20, (uint32_t) dst->ne[1], (uint32_t) dst->ne[2],(uint32_t) dst->ne[3], nb20, (uint32_t) dst->nb[1] / unit_size, (uint32_t) dst->nb[2] / unit_size, (uint32_t) dst->nb[3] / unit_size,
         0,
         0.0f, 0.0f, op_params[0],
-    });
+    }};
+    ggml_vk_op_f32<vk_op_concat_push_constants>(ctx, subctx, src0, src1, nullptr, nullptr, dst, GGML_OP_CONCAT, std::move(pc));
 }
 
 static void ggml_vk_upscale(ggml_backend_vk_context * ctx, vk_context& subctx, const ggml_tensor * src0, ggml_tensor * dst) {
@@ -17872,12 +17932,7 @@ static bool ggml_backend_vk_device_supports_op(ggml_backend_dev_t dev, const ggm
             return op->src[0]->type == op->src[1]->type && op->src[0]->type == op->type &&
                    (op->src[0]->type == GGML_TYPE_F32 || op->src[0]->type == GGML_TYPE_I32);
         case GGML_OP_CONCAT: {
-            if (op->src[0]->type != op->src[1]->type || op->src[0]->type != op->type) {
-                return false;
-            }
-            const size_t type_size = ggml_type_size(op->type);
-            return ggml_blck_size(op->type) == 1 &&
-                   (type_size == 1 || type_size == 2 || type_size == 4 || type_size == 8);
+            return ggml_vk_concat_supported(op->src[0], op->src[1], op);
         }
         case GGML_OP_ADD1:
             return (op->src[0]->type == GGML_TYPE_F32 && op->src[1]->type == GGML_TYPE_F32)

From b4ca032ae3729516943884786de4ae39fba0bbca Mon Sep 17 00:00:00 2001
From: Xuan-Son Nguyen <son@huggingface.co>
Date: Thu, 30 Jul 2026 21:40:38 +0200
Subject: [PATCH 19/30] server: support inp embd to generate next token
 (#26313)

* server: support embd for sampled token

* fix ~server_batch()
---
 tools/server/server-context.cpp | 77 ++++++++++++++++++++++++++++-----
 1 file changed, 66 insertions(+), 11 deletions(-)

diff --git a/tools/server/server-context.cpp b/tools/server/server-context.cpp
index 749bd9aac..0f7ada435 100644
--- a/tools/server/server-context.cpp
+++ b/tools/server/server-context.cpp
@@ -78,31 +78,41 @@ struct server_batch {
     };
     std::vector<token> tokens;
     int32_t n_tokens_alloc = 0;
+    int32_t n_embd = 0;
 
     // track if given slot can be batched with slots already in the batch
     server_slot * slot_batched = nullptr;
 
+    // in embd mode, we temporarily swap out the tokens arr and restore it on clear()
+    bool has_embd = false;
+    llama_token * tokens_ptr = nullptr;
+    std::vector<float> embd;
+
     float  alora_scale       = -1.0f;
     size_t alora_disabled_id = 0;
 
     server_batch() {
-        batch.token = nullptr; // sentinel: uninitialized batch
+        batch.pos = nullptr; // sentinel: uninitialized batch
     }
 
     ~server_batch() {
-        if (batch.token != nullptr) {
+        if (batch.pos != nullptr) {
+            clear();
             llama_batch_free(batch);
         }
     }
 
-    void init(int32_t n_tokens_alloc) {
+    void init(int32_t n_tokens_alloc, int32_t n_embd) {
         this->n_tokens_alloc = n_tokens_alloc;
+        this->n_embd = n_embd;
         batch = llama_batch_init(n_tokens_alloc, 0, 1);
+        tokens_ptr = batch.token;
         tokens.reserve(n_tokens_alloc);
     }
 
     bool add(int32_t id_slot, llama_token token, llama_pos pos, bool output) {
-        GGML_ASSERT(batch.token != nullptr);
+        GGML_ASSERT(!has_embd); // cannot mix tokens + embd in same batch
+        GGML_ASSERT(batch.pos != nullptr);
         if ((int32_t)tokens.size() >= n_tokens_alloc) {
             return false;
         }
@@ -110,13 +120,30 @@ struct server_batch {
         return true;
     }
 
+    bool add(int32_t id_slot, const std::vector<float> & embd_in, llama_pos pos, bool output) {
+        GGML_ASSERT(batch.pos != nullptr);
+        if ((int32_t)tokens.size() >= n_tokens_alloc) {
+            return false;
+        }
+        tokens.push_back({ id_slot, LLAMA_TOKEN_NULL, pos, output });
+        has_embd = true;
+        embd.insert(embd.end(), embd_in.begin(), embd_in.end());
+        return true;
+    }
+
     void clear() {
         tokens.clear();
+        embd.clear();
         common_batch_clear(batch);
         slot_batched      = nullptr;
         alora_scale       = -1.0f;
         alora_disabled_id = 0;
         batch_rendered    = false;
+        has_embd          = false;
+        if (batch.token == nullptr) {
+            batch.token = tokens_ptr;
+            batch.embd  = nullptr;
+        }
     }
 
     int32_t size() const {
@@ -129,25 +156,33 @@ struct server_batch {
     }
 
     void render() {
-        GGML_ASSERT(batch.token != nullptr);
+        GGML_ASSERT(!batch_rendered);
+        GGML_ASSERT(batch.pos != nullptr);
         common_batch_clear(batch);
         for (int32_t i = 0; i < size(); i++) {
             const auto & t = tokens[i];
             common_batch_add(batch, t.token, t.pos, { t.id_slot }, t.output);
         }
+        if (has_embd) {
+            batch.token = nullptr; // will be restored on clear()
+            batch.embd  = embd.data();
+        }
         batch_rendered = true;
     }
 
     llama_batch get_view(int32_t off, int32_t n_tokens) const {
-        GGML_ASSERT(batch.token != nullptr);
+        GGML_ASSERT(batch.pos != nullptr);
         GGML_ASSERT(batch_rendered);
         GGML_ASSERT(off >= 0 && off < size());
         GGML_ASSERT(n_tokens > 0 && off + n_tokens <= size());
 
+        auto * token = batch.token ? batch.token + off          : nullptr;
+        auto * embd  = batch.embd  ? batch.embd  + off * n_embd : nullptr;
+
         llama_batch view = {
             n_tokens,
-            batch.token    + off,
-            nullptr,
+            token,
+            embd,
             batch.pos      + off,
             batch.n_seq_id + off,
             batch.seq_id   + off,
@@ -270,6 +305,10 @@ struct server_slot {
 
     llama_token sampled; // in speculative mode, this is the last accepted token
 
+    // for TTS models, this is the embd generated from prev step, decode this to generate next hidden state
+    // corresponding to one token position (size = n_embd)
+    std::vector<float> inp_embd;
+
     // stats
     size_t n_sent_text = 0; // number of sent text character
 
@@ -378,7 +417,9 @@ struct server_slot {
     bool can_batch_with(server_slot & other_slot) const {
         GGML_ASSERT(task);
 
-        return task->type == other_slot.task->type && are_lora_equal(lora, other_slot.lora);
+        return task->type == other_slot.task->type
+            && inp_embd.size() == other_slot.inp_embd.size()
+            && are_lora_equal(lora, other_slot.lora);
     }
 
     bool has_budget(const common_params & global_params) {
@@ -444,7 +485,11 @@ struct server_slot {
             // no speculative decoding
             i_batch = batch.size();
 
-            add_ok &= batch.add(id, sampled, prompt.tokens.pos_next(), true);
+            if (!inp_embd.empty()) {
+                add_ok &= batch.add(id, inp_embd, prompt.tokens.pos_next(), true);
+            } else {
+                add_ok &= batch.add(id, sampled, prompt.tokens.pos_next(), true);
+            }
 
             SLT_DBG(*this, "slot decode token, id=%d, n_ctx = %d, n_tokens = %d, truncated = %d\n",
                     sampled, n_ctx, prompt.n_tokens(), truncated);
@@ -1334,7 +1379,8 @@ private:
         // note that n_batch can be > n_ctx (e.g. for non-causal attention models such as BERT where the KV cache is not used)
         {
             const int32_t n_batch = llama_n_batch(ctx_tgt);
-            batch.init(std::max(n_batch, params_base.n_parallel));
+            const int32_t n_embd  = llama_model_n_embd_inp(model_tgt);
+            batch.init(std::max(n_batch, params_base.n_parallel), n_embd);
         }
 
         if (params_base.cache_ram_mib != 0) {
@@ -3578,6 +3624,15 @@ private:
             n_empty_consecutive = 0;
         }
 
+        // TODO @ngxson : dft model may have different n_embd than the tgt model, so we check & reject if that's the case
+        // this case is not currently used by any models, but may need to be supported in the future
+        if (spec && batch.has_embd) {
+            if (llama_model_n_embd_inp(model_dft) != llama_model_n_embd_inp(model_tgt)) {
+                SRV_ERR("%s", "unsupported batch.has_embd + spec case\n");
+                throw std::runtime_error("unsupported batch.has_embd + spec case");
+            }
+        }
+
         const int ret = llama_decode(ctx_tgt, batch_view);
 
         metrics.on_decoded(slots);

From 5f55650a78f92aff4d48d671423e888fac0469ff Mon Sep 17 00:00:00 2001
From: Xuan-Son Nguyen <son@huggingface.co>
Date: Thu, 30 Jul 2026 21:59:49 +0200
Subject: [PATCH 20/30] mtmd: add lanczos resize method [no release] (#26341)

---
 tools/mtmd/clip-model.h   |  2 +-
 tools/mtmd/mtmd-image.cpp | 60 ++++++++++++++++++++++++++++++++-------
 2 files changed, 51 insertions(+), 11 deletions(-)

diff --git a/tools/mtmd/clip-model.h b/tools/mtmd/clip-model.h
index 146eabce2..fec2b0180 100644
--- a/tools/mtmd/clip-model.h
+++ b/tools/mtmd/clip-model.h
@@ -33,7 +33,7 @@ enum resize_algo {
     RESIZE_ALGO_BILINEAR, // stretch to target resolution
     RESIZE_ALGO_BICUBIC, // center-crop when aspect ratio doesn't match
     RESIZE_ALGO_BICUBIC_PILLOW,
-    // RESIZE_ALGO_LANCZOS, // TODO
+    RESIZE_ALGO_LANCZOS,
 };
 
 // Padding style for img_tool::resize
diff --git a/tools/mtmd/mtmd-image.cpp b/tools/mtmd/mtmd-image.cpp
index 36cd463b2..72d35fce6 100644
--- a/tools/mtmd/mtmd-image.cpp
+++ b/tools/mtmd/mtmd-image.cpp
@@ -68,6 +68,9 @@ struct img_tool {
                 case RESIZE_ALGO_BICUBIC_PILLOW:
                     resize_bicubic_pillow(src, dst, target_resolution.width, target_resolution.height);
                     break;
+                case RESIZE_ALGO_LANCZOS:
+                    resize_lanczos_pillow(src, dst, target_resolution.width, target_resolution.height);
+                    break;
                 default:
                     throw std::runtime_error("Unsupported resize algorithm");
             }
@@ -97,6 +100,9 @@ struct img_tool {
                 case RESIZE_ALGO_BICUBIC_PILLOW:
                     resize_bicubic_pillow(src, resized_image, new_width, new_height);
                     break;
+                case RESIZE_ALGO_LANCZOS:
+                    resize_lanczos_pillow(src, resized_image, new_width, new_height);
+                    break;
                 default:
                     throw std::runtime_error("Unsupported resize algorithm");
             }
@@ -337,22 +343,50 @@ private:
         }
     }
 
-    // Bicubic resize function using Pillow's ImagingResample algorithm
+    // Pillow-compatible separable resampling (Bicubic and Lanczos)
     // Adapted from https://github.com/python-pillow/Pillow/blob/main/src/libImaging/Resample.c
     //
-    // Key Difference with resize_bicubic:
-    // 1. Uses separable filtering: horizontal pass followed by vertical pass
+    // Key properties:
+    // 1. Separable filtering: horizontal pass followed by vertical pass
     // 2. Pre-computes normalized filter coefficients for each output pixel
-    // 3. Applies convolution using fixed-point integer arithmetic for performance
+    // 3. Fixed-point integer arithmetic (22 fractional bits) for speed and determinism
     static bool resize_bicubic_pillow(const clip_image_u8 & img, clip_image_u8 & dst, int target_width, int target_height) {
+        return resize_pillow(img, dst, target_width, target_height, /*use_lanczos=*/false);
+    }
+
+    // Lanczos-3 (support radius 3), matches Pillow's Image.LANCZOS
+    static bool resize_lanczos_pillow(const clip_image_u8 & img, clip_image_u8 & dst, int target_width, int target_height) {
+        return resize_pillow(img, dst, target_width, target_height, /*use_lanczos=*/true);
+    }
+
+    static bool resize_pillow(
+            const clip_image_u8 & img,
+            clip_image_u8 & dst,
+            int target_width,
+            int target_height,
+            bool use_lanczos) {
         // Fixed-point precision: 22 bits = 32 (int32_t) - 8 (uint8_t pixels) - 2 (headroom for accumulation)
         // This allows encoding fractional weights as integers: weight * 2^22
         const int PRECISION_BITS = 32 - 8 - 2;
 
-        // Bicubic filter function with a = -0.5 (Note that GGML/PyTorch takes a = -0.75)
+        // Resample filter: Lanczos-3 (support [-3, 3]) or bicubic with a = -0.5 (support [-2, 2])
+        // Note: GGML/PyTorch bicubic uses a = -0.75, Pillow uses a = -0.5
         // Returns filter weight for distance x from pixel center
-        // Support: [-2, 2], meaning the filter influences pixels within 2 units of distance
-        auto bicubic_filter = [](double x) -> double {
+        auto resample_filter = [use_lanczos](double x) -> double {
+            if (use_lanczos) {
+                if (-3.0 <= x && x < 3.0) {
+                    auto sinc = [](double v) {
+                        if (v == 0.0) {
+                            return 1.0;
+                        }
+                        const double pi_v = v * 3.141592653589793238462643383279502884;
+                        return std::sin(pi_v) / pi_v;
+                    };
+                    return sinc(x) * sinc(x / 3.0);
+                }
+                return 0.0;
+            }
+
             constexpr double a = -0.5;
             if (x < 0.0) {
                 x = -x;
@@ -366,8 +400,8 @@ private:
             return 0.0;  // Zero outside [-2, 2]
         };
 
-        // Filter support radius: bicubic extends 2 pixels in each direction
-        constexpr double filter_support = 2.0;
+        // Filter support radius: 2 for bicubic, 3 for lanczos
+        const double filter_support = use_lanczos ? 3.0 : 2.0;
 
         // Clipping function for 8-bit values
         auto clip8 = [](int val) -> uint8_t {
@@ -434,7 +468,7 @@ private:
                 // Compute filter weights for each contributing input pixel
                 for (x = 0; x < xmax; x++) {
                     // Distance from input pixel center to output pixel center in input space
-                    double w = bicubic_filter((x + xmin - center + 0.5) * ss);
+                    double w = resample_filter((x + xmin - center + 0.5) * ss);
                     pre_weights[xx * ksize + x] = w;
                     ww += w;  // Accumulate for normalization
                 }
@@ -463,6 +497,12 @@ private:
             const double fxp_scale = std::ldexp(1.0, PRECISION_BITS); // 1.0 * 2^PRECISION_BITS
 
             for (int i = 0; i < outSize * ksize; i++) {
+                if (use_lanczos) {
+                    // Pillow adds +/- 0.5 then truncates toward zero; std::round would round twice
+                    const double rounded = pre_weights[i] * fxp_scale + (pre_weights[i] < 0 ? -0.5 : 0.5);
+                    weights[i] = static_cast<int32_t>(rounded);
+                    continue;
+                }
                 double tmp_val = pre_weights[i] * fxp_scale;
                 if (pre_weights[i] < 0) {
                     tmp_val -= 0.5;

From 8f4646a63ee29f2e0ab971b0290b141938769762 Mon Sep 17 00:00:00 2001
From: Masashi Yoshimura <yoshimura.masashi.frbs@gmail.com>
Date: Fri, 31 Jul 2026 15:08:40 +0900
Subject: [PATCH 21/30] ggml-webgpu: improve flash_attn_vec for quantized KV at
 long contexts (#25956)

* improve fa of quantized kv cache

* Fix some bugs and some comments.

* fix v type check and some comments

* Fix build error caused by rebasing

* editorconfig checking pass
---
 .../ggml-webgpu/ggml-webgpu-shader-lib.hpp    |  75 ++++---
 ggml/src/ggml-webgpu/ggml-webgpu.cpp          |  10 +-
 .../wgsl-shaders/common_decls.tmpl            |  51 ++++-
 .../ggml-webgpu/wgsl-shaders/flash_attn.wgsl  |  27 ++-
 .../flash_attn_quant_staging.tmpl             |  45 +----
 .../wgsl-shaders/flash_attn_tile.wgsl         |   9 +-
 .../wgsl-shaders/flash_attn_vec_split.wgsl    | 186 ++++++++++++++----
 7 files changed, 264 insertions(+), 139 deletions(-)

diff --git a/ggml/src/ggml-webgpu/ggml-webgpu-shader-lib.hpp b/ggml/src/ggml-webgpu/ggml-webgpu-shader-lib.hpp
index 99d775c57..babaddb65 100644
--- a/ggml/src/ggml-webgpu/ggml-webgpu-shader-lib.hpp
+++ b/ggml/src/ggml-webgpu/ggml-webgpu-shader-lib.hpp
@@ -591,7 +591,8 @@ struct ggml_webgpu_flash_attn_common_pipeline_key {
     ggml_type dst_type;
     uint32_t  head_dim_qk;
     uint32_t  head_dim_v;
-    bool      kv_direct;
+    bool      k_direct;
+    bool      v_direct;
     bool      kv_overlap;
     bool      has_mask;
     bool      has_sinks;
@@ -600,8 +601,9 @@ struct ggml_webgpu_flash_attn_common_pipeline_key {
     bool operator==(const ggml_webgpu_flash_attn_common_pipeline_key & other) const {
         return q_type == other.q_type && k_type == other.k_type && v_type == other.v_type &&
                dst_type == other.dst_type && head_dim_qk == other.head_dim_qk && head_dim_v == other.head_dim_v &&
-               kv_direct == other.kv_direct && kv_overlap == other.kv_overlap && has_mask == other.has_mask &&
-               has_sinks == other.has_sinks && uses_logit_softcap == other.uses_logit_softcap;
+               k_direct == other.k_direct && v_direct == other.v_direct && kv_overlap == other.kv_overlap &&
+               has_mask == other.has_mask && has_sinks == other.has_sinks &&
+               uses_logit_softcap == other.uses_logit_softcap;
     }
 };
 
@@ -613,7 +615,8 @@ inline void ggml_webgpu_flash_attn_hash_common_pipeline_key(size_t &
     ggml_webgpu_hash_combine(seed, key.dst_type);
     ggml_webgpu_hash_combine(seed, key.head_dim_qk);
     ggml_webgpu_hash_combine(seed, key.head_dim_v);
-    ggml_webgpu_hash_combine(seed, key.kv_direct);
+    ggml_webgpu_hash_combine(seed, key.k_direct);
+    ggml_webgpu_hash_combine(seed, key.v_direct);
     ggml_webgpu_hash_combine(seed, key.kv_overlap);
     ggml_webgpu_hash_combine(seed, key.has_mask);
     ggml_webgpu_hash_combine(seed, key.has_sinks);
@@ -687,12 +690,13 @@ inline bool ggml_webgpu_flash_attn_float_vec4_aligned(const ggml_tensor * K,
            ggml_webgpu_flash_attn_float_vec4_aligned(V, storage_offset_alignment);
 }
 
-inline bool ggml_webgpu_flash_attn_kv_direct(const ggml_tensor * Q,
-                                             const ggml_tensor * K,
-                                             const ggml_tensor * V,
-                                             uint32_t            kv_direct_align) {
-    return K->type == GGML_TYPE_F16 && V->type == GGML_TYPE_F16 && (Q->ne[0] % kv_direct_align == 0) &&
-           (K->ne[1] % GGML_WEBGPU_KV_SEQ_PAD == 0);
+inline bool ggml_webgpu_flash_attn_k_direct(const ggml_tensor * Q, const ggml_tensor * K, uint32_t kv_direct_align) {
+    return (K->type == GGML_TYPE_F16 || K->type == GGML_TYPE_Q8_0 || K->type == GGML_TYPE_Q4_0) &&
+           (Q->ne[0] % kv_direct_align == 0) && (K->ne[1] % GGML_WEBGPU_KV_SEQ_PAD == 0);
+}
+
+inline bool ggml_webgpu_flash_attn_v_direct(const ggml_tensor * Q, const ggml_tensor * V, uint32_t kv_direct_align) {
+    return ggml_webgpu_flash_attn_k_direct(Q, V, kv_direct_align);
 }
 
 inline ggml_webgpu_flash_attn_common_pipeline_key ggml_webgpu_flash_attn_make_common_pipeline_key(
@@ -706,10 +710,11 @@ inline ggml_webgpu_flash_attn_common_pipeline_key ggml_webgpu_flash_attn_make_co
     key.dst_type                                   = context.dst->type;
     key.head_dim_qk                                = (uint32_t) context.src0->ne[0];
     key.head_dim_v                                 = (uint32_t) context.src2->ne[0];
-    key.kv_direct  = ggml_webgpu_flash_attn_kv_direct(context.src0, context.src1, context.src2, kv_direct_align);
-    key.kv_overlap = kv_overlap;
-    key.has_mask   = context.src3 != nullptr;
-    key.has_sinks  = context.src4 != nullptr;
+    key.k_direct           = ggml_webgpu_flash_attn_k_direct(context.src0, context.src1, kv_direct_align);
+    key.v_direct           = ggml_webgpu_flash_attn_v_direct(context.src0, context.src2, kv_direct_align);
+    key.kv_overlap         = kv_overlap;
+    key.has_mask           = context.src3 != nullptr;
+    key.has_sinks          = context.src4 != nullptr;
     key.uses_logit_softcap = ggml_get_op_params_f32(context.dst, 2) != 0.0f;
     return key;
 }
@@ -794,9 +799,13 @@ inline std::vector<std::string> ggml_webgpu_flash_attn_common_defines(
         defines.push_back("LOGIT_SOFTCAP");
         variant += "_lgsc";
     }
-    if (key.kv_direct) {
-        defines.push_back("KV_DIRECT");
-        variant += "_kvdirect";
+    if (key.k_direct) {
+        defines.push_back("K_DIRECT");
+        variant += "_k_direct";
+    }
+    if (key.v_direct) {
+        defines.push_back("V_DIRECT");
+        variant += "_v_direct";
     }
     if (key.kv_overlap) {
         defines.push_back("KV_OVERLAP");
@@ -815,6 +824,12 @@ inline std::vector<std::string> ggml_webgpu_flash_attn_common_defines(
 
     if (ggml_is_quantized(key.k_type) || ggml_is_quantized(key.v_type)) {
         defines.push_back("U32_DEQUANT_HELPERS");
+        if (ggml_is_quantized(key.k_type)) {
+            defines.push_back("LOADERS_QUANTIZED_K");
+        }
+        if (ggml_is_quantized(key.v_type)) {
+            defines.push_back("LOADERS_QUANTIZED_V");
+        }
     }
 
     return defines;
@@ -2792,12 +2807,14 @@ class ggml_webgpu_shader_lib {
         ggml_webgpu_flash_attn_pipeline_key key = {};
         key.common                              = ggml_webgpu_flash_attn_make_common_pipeline_key(
             context, decisions.use_sg_matrix ? context.sg_mat_k : 1u, kv_overlap);
-        key.common.kv_direct = decisions.use_sg_matrix && key.common.kv_direct;
-        key.use_sg_matrix    = decisions.use_sg_matrix;
+        key.common.k_direct &= decisions.use_sg_matrix && key.common.k_type == GGML_TYPE_F16;
+        key.common.v_direct &= decisions.use_sg_matrix && key.common.v_type == GGML_TYPE_F16;
+        key.use_sg_matrix = decisions.use_sg_matrix;
 
         const uint32_t max_kv_tile = ggml_webgpu_flash_attn_max_kv_tile(
             context.wg_mem_limit_bytes, decisions.q_tile, decisions.use_sg_matrix ? context.sg_mat_n : 1u,
-            key.common.head_dim_qk, key.common.head_dim_v, key.common.has_mask, key.common.kv_direct);
+            key.common.head_dim_qk, key.common.head_dim_v, key.common.has_mask,
+            key.common.k_direct || key.common.v_direct);
         GGML_ASSERT(max_kv_tile > 0);
 
         decisions.kv_tile = decisions.use_sg_matrix ?
@@ -2809,7 +2826,7 @@ class ggml_webgpu_shader_lib {
                 std::min(context.max_wg_size, std::max(GGML_WEBGPU_FLASH_ATTN_PREFERRED_WG_SIZE,
                                                        GGML_WEBGPU_FLASH_ATTN_TILE_Q_TILE * context.max_subgroup_size));
 
-        if (key.common.kv_direct) {
+        if (key.common.k_direct || key.common.v_direct) {
             decisions.kv_tile = std::min(decisions.kv_tile, GGML_WEBGPU_KV_SEQ_PAD);
             while (GGML_WEBGPU_KV_SEQ_PAD % decisions.kv_tile != 0) {
                 decisions.kv_tile -= decisions.use_sg_matrix ? context.sg_mat_n : context.min_subgroup_size;
@@ -2856,9 +2873,9 @@ class ggml_webgpu_shader_lib {
         }
 
         ggml_webgpu_flash_attn_vec_decisions decisions = {};
-        decisions.kv_tile =
-            ggml_webgpu_flash_attn_get_vec_kv_tile(context.wg_mem_limit_bytes, key.common.head_dim_qk,
-                                                   key.common.head_dim_v, key.common.has_mask, key.common.kv_direct);
+        decisions.kv_tile = ggml_webgpu_flash_attn_get_vec_kv_tile(context.wg_mem_limit_bytes, key.common.head_dim_qk,
+                                                                   key.common.head_dim_v, key.common.has_mask,
+                                                                   key.common.k_direct || key.common.v_direct);
         decisions.wg_size = context.max_subgroup_size;
 
         std::string              variant = "flash_attn_vec";
@@ -2870,12 +2887,10 @@ class ggml_webgpu_shader_lib {
             variant += "_mask_blk";
         }
 
-        uint32_t d_split = context.min_subgroup_size;
-        if (key.common.k_type == GGML_TYPE_F16 && key.common.v_type == GGML_TYPE_F16) {
-            const uint32_t D     = key.common.head_dim_qk | key.common.head_dim_v;
-            const uint32_t D_lsb = D & (~(D - 1u));
-            d_split              = std::min(std::min(context.min_subgroup_size, 4u), std::max(D_lsb / 4u, 1u));
-        }
+        uint32_t       d_split = context.min_subgroup_size;
+        const uint32_t D       = key.common.head_dim_qk | key.common.head_dim_v;
+        const uint32_t D_lsb   = D & (~(D - 1u));
+        d_split                = std::min(std::min(context.min_subgroup_size, 4u), std::max(D_lsb / 4u, 1u));
 
         defines.push_back(std::string("D_SPLIT=") + std::to_string(d_split));
         variant += "_dsplit" + std::to_string(d_split);
diff --git a/ggml/src/ggml-webgpu/ggml-webgpu.cpp b/ggml/src/ggml-webgpu/ggml-webgpu.cpp
index 2add5da0b..370f05dfe 100644
--- a/ggml/src/ggml-webgpu/ggml-webgpu.cpp
+++ b/ggml/src/ggml-webgpu/ggml-webgpu.cpp
@@ -3839,7 +3839,8 @@ static size_t ggml_backend_webgpu_buffer_type_get_alloc_size(ggml_backend_buffer
                 const auto &        capabilities = ctx->webgpu_global_ctx->capabilities;
                 if (ggml_webgpu_flash_attn_use_vec_path(ctx->webgpu_global_ctx, Q, K, V)) {
                     const bool kv_direct =
-                        ggml_webgpu_flash_attn_kv_direct(Q, K, V, GGML_WEBGPU_FLASH_ATTN_TILE_KV_VEC_WIDTH);
+                        ggml_webgpu_flash_attn_k_direct(Q, K, GGML_WEBGPU_FLASH_ATTN_TILE_KV_VEC_WIDTH) ||
+                        ggml_webgpu_flash_attn_v_direct(Q, V, GGML_WEBGPU_FLASH_ATTN_TILE_KV_VEC_WIDTH);
                     const uint32_t kv_tile = ggml_webgpu_flash_attn_get_vec_kv_tile(
                         capabilities.limits.maxComputeWorkgroupStorageSize, (uint32_t) Q->ne[0], (uint32_t) V->ne[0],
                         mask != nullptr, kv_direct);
@@ -4448,9 +4449,10 @@ static bool ggml_backend_webgpu_device_supports_op(ggml_backend_dev_t dev, const
                 const uint32_t q_tile =
                     use_subgroup_matrix ? capabilities.sg_mat_m : GGML_WEBGPU_FLASH_ATTN_TILE_Q_TILE;
                 const uint32_t kv_granularity = use_subgroup_matrix ? capabilities.sg_mat_n : 1u;
-                const bool kv_direct = use_subgroup_matrix ?
-                                           ggml_webgpu_flash_attn_kv_direct(src0, src1, src2, capabilities.sg_mat_k) :
-                                           false;
+                const bool     kv_direct = use_subgroup_matrix ?
+                                               ggml_webgpu_flash_attn_k_direct(src0, src1, capabilities.sg_mat_k) ||
+                                                   ggml_webgpu_flash_attn_v_direct(src0, src2, capabilities.sg_mat_k) :
+                                               false;
                 const uint32_t max_kv_tile = ggml_webgpu_flash_attn_max_kv_tile(
                     capabilities.limits.maxComputeWorkgroupStorageSize, q_tile, kv_granularity, (uint32_t) src0->ne[0],
                     (uint32_t) src2->ne[0], op->src[3] != nullptr, kv_direct);
diff --git a/ggml/src/ggml-webgpu/wgsl-shaders/common_decls.tmpl b/ggml/src/ggml-webgpu/wgsl-shaders/common_decls.tmpl
index 6634fbd65..b0cf2853e 100644
--- a/ggml/src/ggml-webgpu/wgsl-shaders/common_decls.tmpl
+++ b/ggml/src/ggml-webgpu/wgsl-shaders/common_decls.tmpl
@@ -9,6 +9,12 @@ fn get_byte_i32(value: u32, index: u32) -> i32 {
 #endif
 
 #ifdef U32_DEQUANT_HELPERS
+
+fn f16_from_u16(bits: u32) -> f16 {
+    let packed = unpack2x16float(bits);
+    return f16(packed[0]);
+}
+
 #ifdef DECLARE_BYTE_LOADERS_SRC
 fn load_u16_at_src(byte_offset: u32) -> u32 {
     let word = src[byte_offset / 4u];
@@ -36,7 +42,7 @@ fn load_f16_as_f32_at_src(byte_offset: u32) -> f32 {
     let d_bits = (word >> shift) & 0xFFFFu;
     return unpack2x16float(d_bits)[0];
 }
-#endif
+#endif // DECLARE_BYTE_LOADERS_SRC
 
 #ifdef DECLARE_BYTE_LOADERS_SRC0
 fn load_u16_at_src0(byte_offset: u32) -> u32 {
@@ -72,8 +78,47 @@ fn load_f16_as_f32_at_src0(byte_offset: u32) -> f32 {
     let d_bits = (word >> shift) & 0xFFFFu;
     return unpack2x16float(d_bits)[0];
 }
-#endif
-#endif
+#endif // DECLARE_BYTE_LOADERS_SRC0
+
+#ifdef LOADERS_QUANTIZED_K
+fn load_k_u16_at(byte_offset: u32) -> u32 {
+    let word = K[byte_offset / 4u];
+    let shift = (byte_offset & 2u) * 8u;
+    return (word >> shift) & 0xFFFFu;
+}
+
+fn load_k_u32_at(byte_offset: u32) -> u32 {
+    let word_idx = byte_offset / 4u;
+    let shift = (byte_offset & 3u) * 8u;
+    let lo = K[word_idx];
+    if (shift == 0u) {
+        return lo;
+    }
+    let hi = K[word_idx + 1u];
+    return (lo >> shift) | (hi << (32u - shift));
+}
+#endif // LOADERS_QUANTIZED_K
+
+#ifdef LOADERS_QUANTIZED_V
+fn load_v_u16_at(byte_offset: u32) -> u32 {
+    let word = V[byte_offset / 4u];
+    let shift = (byte_offset & 2u) * 8u;
+    return (word >> shift) & 0xFFFFu;
+}
+
+fn load_v_u32_at(byte_offset: u32) -> u32 {
+    let word_idx = byte_offset / 4u;
+    let shift = (byte_offset & 3u) * 8u;
+    let lo = V[word_idx];
+    if (shift == 0u) {
+        return lo;
+    }
+    let hi = V[word_idx + 1u];
+    return (lo >> shift) | (hi << (32u - shift));
+}
+#endif // LOADERS_QUANTIZED_V
+
+#endif // U32_DEQUANT_HELPERS
 
 
 
diff --git a/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn.wgsl b/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn.wgsl
index 9767ca3d7..75f33e68a 100644
--- a/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn.wgsl
+++ b/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn.wgsl
@@ -138,7 +138,7 @@ const FLOAT_MIN: f32 = -1.0e9;
 // The number of Q rows processed per workgroup
 var<workgroup> q_shmem: array<f16, Q_TILE * HEAD_DIM_QK>;
 
-#ifndef KV_DIRECT
+#if !defined(K_DIRECT) || !defined(V_DIRECT)
 const kv_shmem_size = KV_TILE * max(HEAD_DIM_QK, HEAD_DIM_V);
 // we can reuse the same shmem for K and V since we only need one at a time
 var<workgroup> kv_shmem: array<f16, kv_shmem_size>;
@@ -183,13 +183,12 @@ fn load_kx4(buf: ptr<storage, array<vec4<K_TYPE>>, read_write>, scalar_index: u3
     return (*buf)[scalar_index >> 2u];
 }
 
-#ifndef KV_DIRECT
+#if !defined(K_DIRECT) || !defined(V_DIRECT)
 #define QUANT_SHMEM kv_shmem
 #define QUANT_OUT_TYPE f16
-#include "quant_inner_loops.tmpl"
 #include "flash_attn_quant_staging.tmpl"
 
-#if !defined(K_Q4_0) && !defined(K_Q8_0)
+#if !defined(K_DIRECT) && !defined(K_Q4_0) && !defined(K_Q8_0)
 fn load_k_tile_block(local_x: u32, kv_count: u32, kv_tile: u32, k_head_offset: u32) {
     for (var elem_idx = local_x; elem_idx < KV_TILE * HEAD_DIM_QK; elem_idx += WG_SIZE) {
         let k_row = elem_idx / HEAD_DIM_QK;
@@ -204,7 +203,7 @@ fn load_k_tile_block(local_x: u32, kv_count: u32, kv_tile: u32, k_head_offset: u
 }
 #endif
 
-#if !defined(V_Q4_0) && !defined(V_Q8_0)
+#if !defined(V_DIRECT) && !defined(V_Q4_0) && !defined(V_Q8_0)
 fn load_v_tile_block(local_x: u32, kv_count: u32, kv_tile: u32, v_head_offset: u32) {
     for (var elem_idx = local_x; elem_idx < KV_TILE * HEAD_DIM_V; elem_idx += WG_SIZE) {
         let v_row = elem_idx / HEAD_DIM_V;
@@ -296,7 +295,7 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
         }
 
       // load k tile into shared memory
-#ifndef KV_DIRECT
+#ifndef K_DIRECT
       load_k_tile_block(local_id.x, kv_count, kv_tile, k_head_offset);
 #endif
 
@@ -306,7 +305,7 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
       // TODO: this loop seems to be the current largest bottleneck
       // this bracket exists to scope the lifetime of variables, reducing register pressure
       {
-#ifdef KV_DIRECT
+#ifdef K_DIRECT
           let k_block_row = kv_tile + subgroup_id * SG_MAT_N;
           var k_global_offset = k_head_offset + k_block_row * params.stride_k1;
 #else
@@ -318,7 +317,7 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
 
               var q_cur = subgroupMatrixLoad<subgroup_matrix_left<f16, SG_MAT_K, SG_MAT_M>>(&q_shmem, 0u, false, HEAD_DIM_QK);
 
-#ifdef KV_DIRECT
+#ifdef K_DIRECT
               var k_cur = subgroupMatrixLoad<subgroup_matrix_right<f16, SG_MAT_N, SG_MAT_K>>(&K, k_global_offset + 0u, true, params.stride_k1);
 #else
               var k_cur = subgroupMatrixLoad<subgroup_matrix_right<f16, SG_MAT_N, SG_MAT_K>>(&kv_shmem, k_block_offset + 0u, true, HEAD_DIM_QK);
@@ -328,7 +327,7 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
               for (; t + 1u < HEAD_DIM_QK / SG_MAT_K; t += 2u) {
                   let h0 = t * SG_MAT_K;
                   var q0 = subgroupMatrixLoad<subgroup_matrix_left<f16, SG_MAT_K, SG_MAT_M>>(&q_shmem, h0, false, HEAD_DIM_QK);
-#ifdef KV_DIRECT
+#ifdef K_DIRECT
                   var k0 = subgroupMatrixLoad<subgroup_matrix_right<f16, SG_MAT_N, SG_MAT_K>>(&K, k_global_offset + h0, true, params.stride_k1);
 #else
                   var k0 = subgroupMatrixLoad<subgroup_matrix_right<f16, SG_MAT_N, SG_MAT_K>>(&kv_shmem, k_block_offset + h0, true, HEAD_DIM_QK);
@@ -339,7 +338,7 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
 
                   let h1 = (t + 1u) * SG_MAT_K;
                   var q1g = subgroupMatrixLoad<subgroup_matrix_left<f16, SG_MAT_K, SG_MAT_M>>(&q_shmem, h1, false, HEAD_DIM_QK);
-#ifdef KV_DIRECT
+#ifdef K_DIRECT
                   var k1g = subgroupMatrixLoad<subgroup_matrix_right<f16, SG_MAT_N, SG_MAT_K>>(&K, k_global_offset + h1, true, params.stride_k1);
 #else
                   var k1g = subgroupMatrixLoad<subgroup_matrix_right<f16, SG_MAT_N, SG_MAT_K>>(&kv_shmem, k_block_offset + h1, true, HEAD_DIM_QK);
@@ -353,7 +352,7 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
               if (t < HEAD_DIM_QK / SG_MAT_K) {
                   let h = t * SG_MAT_K;
                   var qn = subgroupMatrixLoad<subgroup_matrix_left<f16, SG_MAT_K, SG_MAT_M>>(&q_shmem, h, false, HEAD_DIM_QK);
-#ifdef KV_DIRECT
+#ifdef K_DIRECT
                   var kn = subgroupMatrixLoad<subgroup_matrix_right<f16, SG_MAT_N, SG_MAT_K>>(&K, k_global_offset + h, true, params.stride_k1);
 #else
                   var kn = subgroupMatrixLoad<subgroup_matrix_right<f16, SG_MAT_N, SG_MAT_K>>(&kv_shmem, k_block_offset + h, true, HEAD_DIM_QK);
@@ -365,7 +364,7 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
 
               acc = subgroupMatrixMultiplyAccumulate(q_cur, k_cur, acc);
 
-#ifdef KV_DIRECT
+#ifdef K_DIRECT
               k_global_offset += num_subgroups * SG_MAT_N * params.stride_k1;
 #else
               k_block_offset += num_subgroups * SG_MAT_N * HEAD_DIM_QK;
@@ -436,7 +435,7 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
       }
 
       // load v tile into shared memory
-#ifndef KV_DIRECT
+#ifndef V_DIRECT
       load_v_tile_block(local_id.x, kv_count, kv_tile, v_head_offset);
 #endif
 
@@ -464,7 +463,7 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
                   );
 
                   // load V submatrix from global or shared memory
-#ifdef KV_DIRECT
+#ifdef V_DIRECT
                   let v_block_row = kv_tile + kv_block * SG_MAT_N;
                   let v_global_offset = v_head_offset + v_block_row * params.stride_v1 + head_dim_block;
                   var v_sg_mat: subgroup_matrix_right<f16, SG_MAT_N, SG_MAT_K> = subgroupMatrixLoad<subgroup_matrix_right<f16, SG_MAT_N, SG_MAT_K>>(
diff --git a/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_quant_staging.tmpl b/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_quant_staging.tmpl
index 8f41eb7bf..1c23260df 100644
--- a/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_quant_staging.tmpl
+++ b/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_quant_staging.tmpl
@@ -1,3 +1,5 @@
+#include "quant_inner_loops.tmpl"
+
 #define BLOCK_SIZE 32
 #define BLOCKS_K ((HEAD_DIM_QK + BLOCK_SIZE - 1) / BLOCK_SIZE)
 #define BLOCKS_V ((HEAD_DIM_V + BLOCK_SIZE - 1) / BLOCK_SIZE)
@@ -26,49 +28,6 @@
 #define V_BYTES_PER_INNER_LOOP 4u
 #endif
 
-#if defined(K_Q4_0) || defined(K_Q8_0)
-fn load_k_u16_at(byte_offset: u32) -> u32 {
-    let word = K[byte_offset / 4u];
-    let shift = (byte_offset & 2u) * 8u;
-    return (word >> shift) & 0xFFFFu;
-}
-
-fn load_k_u32_at(byte_offset: u32) -> u32 {
-    let word_idx = byte_offset / 4u;
-    let shift = (byte_offset & 3u) * 8u;
-    let lo = K[word_idx];
-    if (shift == 0u) {
-        return lo;
-    }
-    let hi = K[word_idx + 1u];
-    return (lo >> shift) | (hi << (32u - shift));
-}
-#endif
-
-#if defined(V_Q4_0) || defined(V_Q8_0)
-fn load_v_u16_at(byte_offset: u32) -> u32 {
-    let word = V[byte_offset / 4u];
-    let shift = (byte_offset & 2u) * 8u;
-    return (word >> shift) & 0xFFFFu;
-}
-
-fn load_v_u32_at(byte_offset: u32) -> u32 {
-    let word_idx = byte_offset / 4u;
-    let shift = (byte_offset & 3u) * 8u;
-    let lo = V[word_idx];
-    if (shift == 0u) {
-        return lo;
-    }
-    let hi = V[word_idx + 1u];
-    return (lo >> shift) | (hi << (32u - shift));
-}
-#endif
-
-fn f16_from_u16(bits: u32) -> f16 {
-    let packed = unpack2x16float(bits);
-    return f16(packed[0]);
-}
-
 #if defined(K_Q4_0) || defined(K_Q8_0)
 fn load_k_tile_block(local_x: u32, kv_count: u32, kv_tile: u32, k_head_offset: u32) {
     for (var elem_idx = local_x * K_NQ; elem_idx < kv_count * HEAD_DIM_QK; elem_idx += WG_SIZE * K_NQ) {
diff --git a/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_tile.wgsl b/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_tile.wgsl
index e68934113..43f4fe7ca 100644
--- a/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_tile.wgsl
+++ b/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_tile.wgsl
@@ -153,7 +153,6 @@ var<workgroup> p_shmem: array<f16, Q_TILE * KV_TILE>;
 
 #define QUANT_SHMEM kv_shmem
 #define QUANT_OUT_TYPE f16
-#include "quant_inner_loops.tmpl"
 #include "flash_attn_quant_staging.tmpl"
 
 #if !defined(K_Q4_0) && !defined(K_Q8_0)
@@ -270,7 +269,9 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
             local_scores[slot] = FLOAT_MIN;
         }
 
-#ifndef KV_DIRECT
+        // The tile path stages K/V in shared memory so each tile can be reused across
+        // Q_TILE query rows. It therefore does not use the direct path.
+#ifndef K_DIRECT
         load_k_tile_block(local_id.x, kv_count, kv_tile, k_head_offset);
 #endif
 
@@ -333,7 +334,9 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
 
         workgroupBarrier();
 
-#ifndef KV_DIRECT
+        // The tile path stages K/V in shared memory so each tile can be reused across
+        // Q_TILE query rows. It therefore does not use the direct path.
+#ifndef V_DIRECT
         load_v_tile_block(local_id.x, kv_count, kv_tile, v_head_offset);
 #endif
 
diff --git a/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_vec_split.wgsl b/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_vec_split.wgsl
index d51276241..b8e0be90d 100644
--- a/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_vec_split.wgsl
+++ b/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn_vec_split.wgsl
@@ -196,49 +196,35 @@ struct Params {
 
 // Just a very small float value.
 const FLOAT_MIN: f32 = -1.0e9;
+const kv_shmem_size = KV_TILE * max(HEAD_DIM_QK, HEAD_DIM_V);
 
 var<workgroup> q_shmem: array<f32, HEAD_DIM_QK>;
-
-#ifndef KV_DIRECT
-const kv_shmem_size = KV_TILE * max(HEAD_DIM_QK, HEAD_DIM_V);
-// we can reuse the same shmem for K and V since we only need one at a time
-var<workgroup> kv_shmem: array<f32, kv_shmem_size>;
-#endif
-
 var<workgroup> o_shmem: array<f32, HEAD_DIM_V>;
+// note that we reuse the same storage for both since we only need one at a time
+var<workgroup> inter_shmem: array<f32, KV_TILE>;
 
 #ifdef MASK
 // storage for mask values
 var<workgroup> mask_shmem: array<f32, KV_TILE>;
 #endif
 
-// note that we reuse the same storage for both since we only need one at a time
-var<workgroup> inter_shmem: array<f32, KV_TILE>;
-
-// Storage for row max and exp sum during online softmax
-fn calc_softmax_term(kv_idx: u32, slope: f32, has_bias: bool, apply_mask: bool) -> f32 {
-    var v = select(FLOAT_MIN,
-                   inter_shmem[kv_idx] * params.scale,
-                   kv_idx < KV_TILE);
-#ifdef LOGIT_SOFTCAP
-    v = params.logit_softcap * tanh(v);
+#if defined(K_DIRECT) || defined(V_DIRECT)
+// Shared memory for scale factor (d) in quantized K/V. Multiple threads use the same value,
+// so caching it is more efficient, even on the direct path.
+var<workgroup> d_shmem: array<f32, kv_shmem_size / 32>;
 #endif
-#ifdef MASK
-    if (apply_mask) {
-        var mask_val = select(0.0, mask_shmem[kv_idx], kv_idx < KV_TILE);
-        v += select(mask_val, slope * mask_val, has_bias);
-    }
-#endif
-    return v;
-}
 
-#ifndef KV_DIRECT
+// K/V shared memory handling
+#if !defined(K_DIRECT) || !defined(V_DIRECT)
+
+// we can reuse the same shmem for K and V since we only need one at a time
+var<workgroup> kv_shmem: array<f32, kv_shmem_size>;
+
 #define QUANT_SHMEM kv_shmem
 #define QUANT_OUT_TYPE f32
-#include "quant_inner_loops.tmpl"
 #include "flash_attn_quant_staging.tmpl"
 
-#if !defined(K_Q4_0) && !defined(K_Q8_0)
+#if !defined(K_DIRECT) && !defined(K_Q4_0) && !defined(K_Q8_0)
 fn load_k_tile_block(local_x: u32, kv_count: u32, kv_tile: u32, k_head_offset: u32) {
     for (var elem_idx = local_x * 4u; elem_idx < KV_TILE * HEAD_DIM_QK; elem_idx += WG_SIZE * 4u) {
         let k_row = elem_idx / HEAD_DIM_QK;
@@ -256,7 +242,7 @@ fn load_k_tile_block(local_x: u32, kv_count: u32, kv_tile: u32, k_head_offset: u
 }
 #endif
 
-#if !defined(V_Q4_0) && !defined(V_Q8_0)
+#if !defined(V_DIRECT) && !defined(V_Q4_0) && !defined(V_Q8_0)
 fn load_v_tile_block(local_x: u32, kv_count: u32, kv_tile: u32, v_head_offset: u32) {
     for (var elem_idx = local_x * 4u; elem_idx < KV_TILE * HEAD_DIM_V; elem_idx += WG_SIZE * 4u) {
         let v_row = elem_idx / HEAD_DIM_V;
@@ -273,7 +259,24 @@ fn load_v_tile_block(local_x: u32, kv_count: u32, kv_tile: u32, v_head_offset: u
     }
 }
 #endif
+#endif // !defined(K_DIRECT) || !defined(V_DIRECT)
+
+// Storage for row max and exp sum during online softmax
+fn calc_softmax_term(kv_idx: u32, slope: f32, has_bias: bool, apply_mask: bool) -> f32 {
+    var v = select(FLOAT_MIN,
+                   inter_shmem[kv_idx] * params.scale,
+                   kv_idx < KV_TILE);
+#ifdef LOGIT_SOFTCAP
+    v = params.logit_softcap * tanh(v);
 #endif
+#ifdef MASK
+    if (apply_mask) {
+        var mask_val = select(0.0, mask_shmem[kv_idx], kv_idx < KV_TILE);
+        v += select(mask_val, slope * mask_val, has_bias);
+    }
+#endif
+    return v;
+}
 
 @compute @workgroup_size(WG_SIZE)
 fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
@@ -355,12 +358,31 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
             inter_shmem[elem_idx] = 0.0;
         }
 
-      // load k tile into shared memory
-#ifndef KV_DIRECT
-      load_k_tile_block(local_id.x, kv_count, kv_tile, k_head_offset);
+#ifdef K_DIRECT
+      // load only the scale factor (d) from each quantized block into shared memory on the direct path.
+#if defined(K_Q8_0)
+        for (var j = local_id.x * 32; j < KV_TILE * HEAD_DIM_QK; j += WG_SIZE * 32) {
+            let kv_row = kv_tile + j / HEAD_DIM_QK;
+            let block_idx = (j % HEAD_DIM_QK) / 32;
+            let block_byte_base = 34 * (k_head_offset + kv_row * params.stride_k1 + block_idx);
+            let d = f32(f16_from_u16(load_k_u16_at(block_byte_base)));
+            d_shmem[j / 32] = d;
+        }
+#elif defined(K_Q4_0)
+        for (var j = local_id.x * 32; j < KV_TILE * HEAD_DIM_QK; j += WG_SIZE * 32) {
+            let kv_row = kv_tile + j / HEAD_DIM_QK;
+            let block_idx = (j % HEAD_DIM_QK) / 32;
+            let block_byte_base = 18 * (k_head_offset + kv_row * params.stride_k1 + block_idx);
+            let d = f32(f16_from_u16(load_k_u16_at(block_byte_base)));
+            d_shmem[j / 32] = d;
+        }
 #endif
+#else
+      // load k tile into shared memory
+      load_k_tile_block(local_id.x, kv_count, kv_tile, k_head_offset);
+#endif // defined(K_DIRECT)
 
-      workgroupBarrier();
+        workgroupBarrier();
 
       // accumulate q block * k block into registers across the entire KV tile
       if (!skip_tile) {
@@ -381,9 +403,40 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
                             q_shmem[q_off + 1u],
                             q_shmem[q_off + 2u],
                             q_shmem[q_off + 3u]);
-#ifdef KV_DIRECT
+#ifdef K_DIRECT
+#if defined(K_Q8_0)
+                        let kv_row = kv_tile + kv_idx;
+                        let block_idx = (i * 4u) / 32;
+                        let id_in_block = (i * 4u) % 32;
+                        let block_byte_base = 34 * (k_head_offset + kv_row * params.stride_k1 + block_idx);
+                        let q_byte_base = block_byte_base + 2u;
+                        let d = d_shmem[(kv_idx * HEAD_DIM_QK) / 32 + block_idx];
+                        let q8u4 = load_k_u32_at(q_byte_base + id_in_block);
+                        let kv = vec4<f32>(
+                            d * f32(get_byte_i32(q8u4, 0)),
+                            d * f32(get_byte_i32(q8u4, 1)),
+                            d * f32(get_byte_i32(q8u4, 2)),
+                            d * f32(get_byte_i32(q8u4, 3)),
+                        );
+#elif defined(K_Q4_0)
+                        let kv_row = kv_tile + kv_idx;
+                        let block_idx = (i * 4u) / 32;
+                        let id_in_block = (i * 4u) % 32;
+                        let phase = id_in_block / 16;
+                        let block_byte_base = 18 * (k_head_offset + kv_row * params.stride_k1 + block_idx);
+                        let q_byte_base = block_byte_base + 2u;
+                        let d = d_shmem[(kv_idx * HEAD_DIM_QK) / 32 + block_idx];
+                        let q8u4 = load_k_u32_at(q_byte_base + (id_in_block - phase * 16u));
+                        let kv = vec4<f32>(
+                            d * (f32((get_byte(q8u4, 0) >> (phase * 4u)) & 0xFu) - 8.0),
+                            d * (f32((get_byte(q8u4, 1) >> (phase * 4u)) & 0xFu) - 8.0),
+                            d * (f32((get_byte(q8u4, 2) >> (phase * 4u)) & 0xFu) - 8.0),
+                            d * (f32((get_byte(q8u4, 3) >> (phase * 4u)) & 0xFu) - 8.0),
+                        );
+#else
                         let idx = k_head_offset + (kv_tile + kv_idx) * params.stride_k1 + (i * 4u);
                         let kv = vec4<f32>(K[idx >> 2u]);
+#endif
 #else
                         let idx = kv_idx * HEAD_DIM_QK + (i * 4u);
                         let kv = vec4<f32>(
@@ -391,7 +444,7 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
                             kv_shmem[idx + 1u],
                             kv_shmem[idx + 2u],
                             kv_shmem[idx + 3u]);
-#endif
+#endif // defined(K_DIRECT)
                         partial_sum += dot(qv, kv);
                     }
                   }
@@ -473,12 +526,32 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
           }
       }
 
-      // load v tile into shared memory
-#ifndef KV_DIRECT
-      load_v_tile_block(local_id.x, kv_count, kv_tile, v_head_offset);
-#endif
 
-      workgroupBarrier();
+#ifdef V_DIRECT
+      // load only `d` of quantized block into shared memory in the direct path
+#if defined(V_Q8_0)
+        for (var j = local_id.x * 32; j < KV_TILE * HEAD_DIM_V; j += WG_SIZE * 32) {
+            let v_row = kv_tile + j / HEAD_DIM_V;
+            let block_idx = (j % HEAD_DIM_V) / 32;
+            let block_byte_base = 34 * (v_head_offset + v_row * params.stride_v1 + block_idx);
+            let d = f32(f16_from_u16(load_v_u16_at(block_byte_base)));
+            d_shmem[j / 32] = d;
+        }
+#elif defined(V_Q4_0)
+        for (var j = local_id.x * 32; j < KV_TILE * HEAD_DIM_V; j += WG_SIZE * 32) {
+            let v_row = kv_tile + j / HEAD_DIM_V;
+            let block_idx = (j % HEAD_DIM_V) / 32;
+            let block_byte_base = 18 * (v_head_offset + v_row * params.stride_v1 + block_idx);
+            let d = f32(f16_from_u16(load_v_u16_at(block_byte_base)));
+            d_shmem[j / 32] = d;
+        }
+#endif
+#else
+      // load v tile into shared memory
+      load_v_tile_block(local_id.x, kv_count, kv_tile, v_head_offset);
+#endif // V_DIRECT
+
+        workgroupBarrier();
 
       if (!skip_tile) {
           // we have P (KV_TILE) in inter_shmem and V (KV_TILE x head_dim_v) in kv_shmem
@@ -501,9 +574,38 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
                       }
 
                       let p = inter_shmem[kv_idx];
-#ifdef KV_DIRECT
+#ifdef V_DIRECT
+#if defined(V_Q8_0)
+                        let block_idx = (vec_col * 4u) / 32;
+                        let id_in_block = (vec_col * 4u) % 32;
+                        let block_byte_base = 34 * (v_head_offset + v_row * params.stride_v1 + block_idx);
+                        let q_byte_base = block_byte_base + 2u;
+                        let d = d_shmem[(kv_idx * HEAD_DIM_V) / 32 + block_idx];
+                        let q8u4 = load_v_u32_at(q_byte_base + id_in_block);
+                        let v4 = vec4<f32>(
+                            d * f32(get_byte_i32(q8u4, 0)),
+                            d * f32(get_byte_i32(q8u4, 1)),
+                            d * f32(get_byte_i32(q8u4, 2)),
+                            d * f32(get_byte_i32(q8u4, 3)),
+                        );
+#elif defined(V_Q4_0)
+                        let block_idx = (vec_col * 4u) / 32;
+                        let id_in_block = (vec_col * 4u) % 32;
+                        let phase = id_in_block / 16;
+                        let block_byte_base = 18 * (v_head_offset + v_row * params.stride_v1 + block_idx);
+                        let q_byte_base = block_byte_base + 2u;
+                        let d = d_shmem[(kv_idx * HEAD_DIM_V) / 32 + block_idx];
+                        let q8u4 = load_v_u32_at(q_byte_base + (id_in_block - phase * 16u));
+                        let v4 = vec4<f32>(
+                            d * (f32((get_byte(q8u4, 0) >> (phase * 4u)) & 0xFu) - 8.0),
+                            d * (f32((get_byte(q8u4, 1) >> (phase * 4u)) & 0xFu) - 8.0),
+                            d * (f32((get_byte(q8u4, 2) >> (phase * 4u)) & 0xFu) - 8.0),
+                            d * (f32((get_byte(q8u4, 3) >> (phase * 4u)) & 0xFu) - 8.0),
+                        );
+#else
                       let v_idx = v_head_offset + v_row * params.stride_v1 + vec_col * 4u;
                       let v4 = vec4<f32>(V[v_idx >> 2u]);
+#endif
 #else
                       let v_idx = kv_idx * HEAD_DIM_V + vec_col * 4u;
                       let v4 = vec4<f32>(
@@ -511,7 +613,7 @@ fn main(@builtin(workgroup_id) wg_id: vec3<u32>,
                           kv_shmem[v_idx + 1u],
                           kv_shmem[v_idx + 2u],
                           kv_shmem[v_idx + 3u]);
-#endif
+#endif // defined(V_DIRECT)
                       lo += p * v4;
                   }
 

From 1553725965470086976c8a06f31467970c9b2702 Mon Sep 17 00:00:00 2001
From: Titaniumtown <titaniumtown@proton.me>
Date: Thu, 30 Jul 2026 23:17:53 -0700
Subject: [PATCH 22/30] sycl: fuse RMS_NORM + MUL (#26015)

---
 ggml/src/ggml-sycl/backend.hpp   |   1 +
 ggml/src/ggml-sycl/fusion.cpp    |  44 ++++++++++++
 ggml/src/ggml-sycl/fusion.hpp    |  15 ++++
 ggml/src/ggml-sycl/ggml-sycl.cpp |   7 ++
 ggml/src/ggml-sycl/norm.cpp      | 120 ++++++++++++++++++++++++++++++-
 ggml/src/ggml-sycl/norm.hpp      |   2 +
 6 files changed, 187 insertions(+), 2 deletions(-)
 create mode 100644 ggml/src/ggml-sycl/fusion.cpp
 create mode 100644 ggml/src/ggml-sycl/fusion.hpp

diff --git a/ggml/src/ggml-sycl/backend.hpp b/ggml/src/ggml-sycl/backend.hpp
index f299bcf62..51ab6f930 100644
--- a/ggml/src/ggml-sycl/backend.hpp
+++ b/ggml/src/ggml-sycl/backend.hpp
@@ -26,6 +26,7 @@
 #include "dmmv.hpp"
 #include "element_wise.hpp"
 #include "fattn.hpp"
+#include "fusion.hpp"
 #include "gated_delta_net.hpp"
 #include "gla.hpp"
 #include "im2col.hpp"
diff --git a/ggml/src/ggml-sycl/fusion.cpp b/ggml/src/ggml-sycl/fusion.cpp
new file mode 100644
index 000000000..4a6027f39
--- /dev/null
+++ b/ggml/src/ggml-sycl/fusion.cpp
@@ -0,0 +1,44 @@
+#include "fusion.hpp"
+
+bool ggml_sycl_can_fuse(const ggml_cgraph * cgraph, int node_idx, std::initializer_list<enum ggml_op> ops) {
+    if (!g_ggml_sycl_enable_fusion) {
+        return false;
+    }
+
+    if (!ggml_can_fuse(cgraph, node_idx, ops)) {
+        return false;
+    }
+
+    if (ops.size() == 2 && ops.begin()[0] == GGML_OP_RMS_NORM && ops.begin()[1] == GGML_OP_MUL) {
+        const ggml_tensor * rms_norm = cgraph->nodes[node_idx];
+        const ggml_tensor * mul      = cgraph->nodes[node_idx + 1];
+
+        GGML_ASSERT(rms_norm->src[0]->type == GGML_TYPE_F32);
+        GGML_ASSERT(rms_norm->type == GGML_TYPE_F32);
+
+        if (mul->src[0]->type != GGML_TYPE_F32 ||
+            mul->src[1]->type != GGML_TYPE_F32 ||
+            mul->type != GGML_TYPE_F32) {
+            return false;
+        }
+
+        // if rms norm is the B operand, then we don't handle broadcast
+        if (rms_norm == mul->src[1] && !ggml_are_same_shape(mul->src[0], rms_norm)) {
+            return false;
+        }
+
+        const ggml_tensor * mul_w = (mul->src[0] == rms_norm) ? mul->src[1] : mul->src[0];
+        // the fused kernel indexes the weight as mul[col], so it must span ncols contiguously
+        if (mul_w->ne[0] != rms_norm->ne[0] || mul_w->nb[0] != ggml_type_size(mul_w->type)) {
+            return false;
+        }
+
+        if (!ggml_is_contiguous_rows(mul->src[0]) || !ggml_is_contiguous_rows(mul->src[1])) {
+            return false;
+        }
+
+        return true;
+    }
+
+    return false;
+}
diff --git a/ggml/src/ggml-sycl/fusion.hpp b/ggml/src/ggml-sycl/fusion.hpp
new file mode 100644
index 000000000..7d7c79e02
--- /dev/null
+++ b/ggml/src/ggml-sycl/fusion.hpp
@@ -0,0 +1,15 @@
+#ifndef GGML_SYCL_FUSION_HPP
+#define GGML_SYCL_FUSION_HPP
+
+#include <initializer_list>
+
+#include "common.hpp"
+
+// Backend-side fusability test. `ops` names a candidate op sequence starting at cgraph node
+// `node_idx`; the result is true only if ggml considers that subgraph fusable *and* the SYCL
+// kernel which would service it accepts the tensors involved (types, shapes, contiguity).
+//
+// Lives in its own translation unit because it grows a branch per supported op sequence.
+bool ggml_sycl_can_fuse(const ggml_cgraph * cgraph, int node_idx, std::initializer_list<enum ggml_op> ops);
+
+#endif  // GGML_SYCL_FUSION_HPP
diff --git a/ggml/src/ggml-sycl/ggml-sycl.cpp b/ggml/src/ggml-sycl/ggml-sycl.cpp
index 3b807c7cb..49bdc97d2 100644
--- a/ggml/src/ggml-sycl/ggml-sycl.cpp
+++ b/ggml/src/ggml-sycl/ggml-sycl.cpp
@@ -5398,6 +5398,13 @@ static void ggml_backend_sycl_graph_compute_impl(ggml_backend_sycl_context * syc
             }
         }
 #endif
+        if (node->op == GGML_OP_RMS_NORM &&
+            ggml_sycl_can_fuse(cgraph, i, { GGML_OP_RMS_NORM, GGML_OP_MUL })) {
+            ggml_sycl_op_rms_norm_fused(*sycl_ctx, node, cgraph->nodes[i + 1]);
+            i++;
+            continue;
+        }
+
         bool ok = ggml_sycl_compute_forward(*sycl_ctx, node);
         if (!ok) {
             GGML_LOG_ERROR("%s: error: op not supported %s (%s)\n", __func__, node->name, ggml_op_name(node->op));
diff --git a/ggml/src/ggml-sycl/norm.cpp b/ggml/src/ggml-sycl/norm.cpp
index c4472e4bd..682a9f51e 100644
--- a/ggml/src/ggml-sycl/norm.cpp
+++ b/ggml/src/ggml-sycl/norm.cpp
@@ -147,10 +147,13 @@ static void group_norm_f32(const float* x, float* dst, const int group_size, con
     }
 }
 
+template <bool do_multiply = false>
 static void rms_norm_f32(const float* x, float* dst, const int ncols,
     const int64_t src_stride_col, const int64_t src_stride_row, const int64_t src_stride_channel, const int64_t src_stride_sample,
     const int64_t dst_stride_col, const int64_t dst_stride_row, const int64_t dst_stride_channel, const int64_t dst_stride_sample,
-    const float eps, const sycl::nd_item<3>& item_ct1, float* s_sum, int block_size) {
+    const float eps, const sycl::nd_item<3>& item_ct1, float* s_sum, int block_size,
+    const float* mul = nullptr, const int64_t mul_stride_row = 0, const int64_t mul_stride_channel = 0,
+    const int64_t mul_stride_sample = 0, const int mul_nrows = 0, const int mul_nchannels = 0, const int mul_nsamples = 0) {
 
     const int nrows = item_ct1.get_group_range(2);
     const int nchannels = item_ct1.get_group_range(1);
@@ -170,6 +173,12 @@ static void rms_norm_f32(const float* x, float* dst, const int ncols,
     x   += src_offset;
     dst += dst_offset;
 
+    if constexpr (do_multiply) {
+        const int mul_row     = row     % mul_nrows;
+        const int mul_channel = channel % mul_nchannels;
+        const int mul_sample  = sample  % mul_nsamples;
+        mul += mul_sample * mul_stride_sample + mul_channel * mul_stride_channel + mul_row * mul_stride_row;
+    }
 
     float tmp = 0.0f; // partial sum for thread in warp
 
@@ -202,7 +211,11 @@ static void rms_norm_f32(const float* x, float* dst, const int ncols,
     const float scale = sycl::rsqrt(mean + eps);
 
     for (int col = tid; col < ncols; col += block_size) {
-        dst[col * dst_stride_col] = scale * x[col * src_stride_col];
+        if constexpr (do_multiply) {
+            dst[col * dst_stride_col] = scale * x[col * src_stride_col] * mul[col];
+        } else {
+            dst[col * dst_stride_col] = scale * x[col * src_stride_col];
+        }
     }
 }
 
@@ -376,6 +389,49 @@ static void rms_norm_f32_sycl(const float* x, float* dst, const int ncols, const
     }
 }
 
+static void rms_norm_mul_f32_sycl(const float* x, const float* mul, float* dst, const int ncols, const int nrows,
+        const int nchannels, const int nsamples,
+        const int64_t src_stride_col, const int64_t src_stride_row, const int64_t src_stride_channel, const int64_t src_stride_sample,
+        const int64_t dst_stride_col, const int64_t dst_stride_row, const int64_t dst_stride_channel, const int64_t dst_stride_sample,
+        const int64_t mul_stride_row, const int64_t mul_stride_channel, const int64_t mul_stride_sample,
+        const int mul_nrows, const int mul_nchannels, const int mul_nsamples,
+        const float eps, queue_ptr stream, int device) {
+    const sycl::range<3> global_dims(nsamples, nchannels, nrows);
+    if (ncols < 1024) {
+        const sycl::range<3> block_dims(1, 1, WARP_SIZE);
+        stream->submit([&](sycl::handler& cgh) {
+            cgh.parallel_for(
+                sycl::nd_range<3>(global_dims * block_dims, block_dims),
+                [=](sycl::nd_item<3> item_ct1)
+                [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
+                    rms_norm_f32<true>(x, dst, ncols,
+                        src_stride_col, src_stride_row, src_stride_channel, src_stride_sample,
+                        dst_stride_col, dst_stride_row, dst_stride_channel, dst_stride_sample,
+                        eps, item_ct1, nullptr, WARP_SIZE,
+                        mul, mul_stride_row, mul_stride_channel, mul_stride_sample, mul_nrows, mul_nchannels, mul_nsamples);
+                });
+            });
+    }
+    else {
+        const int work_group_size = ggml_sycl_info().max_work_group_sizes[device];
+        assert(work_group_size % (WARP_SIZE * WARP_SIZE) == 0);
+        const sycl::range<3> block_dims(1, 1, work_group_size);
+        stream->submit([&](sycl::handler& cgh) {
+            sycl::local_accessor<float, 1> s_sum_acc_ct1(sycl::range<1>(work_group_size / WARP_SIZE), cgh);
+            cgh.parallel_for(
+                sycl::nd_range<3>(global_dims * block_dims, block_dims),
+                [=](sycl::nd_item<3> item_ct1)
+                [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
+                    rms_norm_f32<true>(x, dst, ncols,
+                        src_stride_col, src_stride_row, src_stride_channel, src_stride_sample,
+                        dst_stride_col, dst_stride_row, dst_stride_channel, dst_stride_sample,
+                        eps, item_ct1, get_pointer(s_sum_acc_ct1), work_group_size,
+                        mul, mul_stride_row, mul_stride_channel, mul_stride_sample, mul_nrows, mul_nchannels, mul_nsamples);
+                });
+            });
+    }
+}
+
 template<int warp_size>
 static void l2_norm_f32_sycl(const float *   x,
                              float *         dst,
@@ -518,6 +574,66 @@ void ggml_sycl_op_rms_norm(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
         ss0, ss1, ss2, ss3, ds0, ds1, ds2, ds3, eps, main_stream, ctx.device);
 }
 
+void ggml_sycl_op_rms_norm_fused(ggml_backend_sycl_context & ctx, ggml_tensor * dst, ggml_tensor * mul_tensor) {
+    const ggml_tensor * rms_norm_src = dst->src[0];
+    float eps = 0.0f;
+    memcpy(&eps, dst->op_params, sizeof(float));
+
+    const float *       src0_dd = static_cast<const float *>(rms_norm_src->data);
+    const float *       mul_dd  = nullptr;
+    const ggml_tensor * mul_src = nullptr;
+    if (mul_tensor->src[0] == dst) {
+        mul_dd  = static_cast<const float *>(mul_tensor->src[1]->data);
+        mul_src = mul_tensor->src[1];
+    } else if (mul_tensor->src[1] == dst) {
+        mul_dd  = static_cast<const float *>(mul_tensor->src[0]->data);
+        mul_src = mul_tensor->src[0];
+    } else {
+        GGML_ASSERT(false);
+    }
+    float * dst_dd = static_cast<float *>(mul_tensor->data);
+
+    dpct::queue_ptr main_stream = ctx.stream();
+    SYCL_CHECK(ggml_sycl_set_device(ctx.device));
+
+    GGML_ASSERT(rms_norm_src->type == GGML_TYPE_F32);
+    GGML_ASSERT(dst->type == GGML_TYPE_F32);
+    GGML_ASSERT(mul_tensor->type == GGML_TYPE_F32);
+    GGML_ASSERT(eps >= 0.0f);
+
+    const int64_t ne00 = rms_norm_src->ne[0];
+    const int64_t ne01 = rms_norm_src->ne[1];
+    const int64_t ne02 = rms_norm_src->ne[2];
+    const int64_t ne03 = rms_norm_src->ne[3];
+
+    const size_t ts0 = ggml_type_size(rms_norm_src->type);
+    GGML_ASSERT(rms_norm_src->nb[0] == ts0);
+    const int64_t s00 = rms_norm_src->nb[0] / ts0;
+    const int64_t s01 = rms_norm_src->nb[1] / ts0;
+    const int64_t s02 = rms_norm_src->nb[2] / ts0;
+    const int64_t s03 = rms_norm_src->nb[3] / ts0;
+
+    const size_t tdst = ggml_type_size(mul_tensor->type);
+    GGML_ASSERT(mul_tensor->nb[0] == tdst);
+    const int64_t d00 = mul_tensor->nb[0] / tdst;
+    const int64_t d01 = mul_tensor->nb[1] / tdst;
+    const int64_t d02 = mul_tensor->nb[2] / tdst;
+    const int64_t d03 = mul_tensor->nb[3] / tdst;
+
+    const size_t ts_mul = ggml_type_size(mul_src->type);
+    GGML_ASSERT(mul_src->nb[0] == ts_mul);
+    const int64_t mul_s01 = mul_src->nb[1] / ts_mul;
+    const int64_t mul_s02 = mul_src->nb[2] / ts_mul;
+    const int64_t mul_s03 = mul_src->nb[3] / ts_mul;
+    const int mul_nrows     = mul_src->ne[1];
+    const int mul_nchannels = mul_src->ne[2];
+    const int mul_nsamples  = mul_src->ne[3];
+
+    rms_norm_mul_f32_sycl(src0_dd, mul_dd, dst_dd, ne00, ne01, ne02, ne03,
+        s00, s01, s02, s03, d00, d01, d02, d03,
+        mul_s01, mul_s02, mul_s03, mul_nrows, mul_nchannels, mul_nsamples, eps, main_stream, ctx.device);
+}
+
 void ggml_sycl_op_rms_norm_back(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
     scope_op_debug_print scope_dbg_print(__func__, dst, /*num_src=*/2);
 
diff --git a/ggml/src/ggml-sycl/norm.hpp b/ggml/src/ggml-sycl/norm.hpp
index 8cb885eb2..51217c421 100644
--- a/ggml/src/ggml-sycl/norm.hpp
+++ b/ggml/src/ggml-sycl/norm.hpp
@@ -19,6 +19,8 @@ void ggml_sycl_op_norm(ggml_backend_sycl_context& ctx, ggml_tensor* dst);
 
 void ggml_sycl_op_rms_norm(ggml_backend_sycl_context& ctx, ggml_tensor* dst);
 
+void ggml_sycl_op_rms_norm_fused(ggml_backend_sycl_context& ctx, ggml_tensor* dst, ggml_tensor* mul);
+
 void ggml_sycl_op_rms_norm_back(ggml_backend_sycl_context& ctx, ggml_tensor* dst);
 
 void ggml_sycl_op_group_norm(ggml_backend_sycl_context& ctx, ggml_tensor* dst);

From a2be61dc87947e069dcdfddcf276abd7d629e924 Mon Sep 17 00:00:00 2001
From: Neo Zhang <zhang.jianyu@outlook.com>
Date: Fri, 31 Jul 2026 14:19:41 +0800
Subject: [PATCH 23/30] [SYCL] Support q2 mul_mat (#26231)

* support q2_0 in mul_mat

* support more q2_0 case
---
 ggml/src/ggml-sycl/convert.cpp    |  4 ++
 ggml/src/ggml-sycl/dequantize.hpp | 22 +++++++++
 ggml/src/ggml-sycl/mmvq.cpp       | 79 +++++++++++++++++++++++++++++++
 ggml/src/ggml-sycl/vecdotq.hpp    | 69 +++++++++++++++++++++++++++
 4 files changed, 174 insertions(+)

diff --git a/ggml/src/ggml-sycl/convert.cpp b/ggml/src/ggml-sycl/convert.cpp
index 060d0aca2..9ec927695 100644
--- a/ggml/src/ggml-sycl/convert.cpp
+++ b/ggml/src/ggml-sycl/convert.cpp
@@ -644,6 +644,8 @@ to_fp16_sycl_t ggml_get_to_fp16_sycl(ggml_type type, ggml_tensor * dst) {
     switch (type) {
         case GGML_TYPE_Q1_0:
             return dequantize_block_sycl<QK1_0, QR1_0, dequantize_q1_0>;
+        case GGML_TYPE_Q2_0:
+            return dequantize_block_sycl<QK2_0, QR2_0, dequantize_q2_0>;
         case GGML_TYPE_Q4_0:
             if (dst->src[0]->extra &&
                 ((ggml_tensor_extra_gpu*)dst->src[0]->extra)->optimized_feature.reorder) {
@@ -728,6 +730,8 @@ to_fp32_sycl_t ggml_get_to_fp32_sycl(ggml_type type, ggml_tensor *dst) {
     switch (type) {
         case GGML_TYPE_Q1_0:
             return dequantize_block_sycl<QK1_0, QR1_0, dequantize_q1_0>;
+        case GGML_TYPE_Q2_0:
+            return dequantize_block_sycl<QK2_0, QR2_0, dequantize_q2_0>;
         case GGML_TYPE_Q4_0:
             if (dst->src[0]->extra &&
                 ((ggml_tensor_extra_gpu*)dst->src[0]->extra)->optimized_feature.reorder) {
diff --git a/ggml/src/ggml-sycl/dequantize.hpp b/ggml/src/ggml-sycl/dequantize.hpp
index 3db55319f..876ba1b44 100644
--- a/ggml/src/ggml-sycl/dequantize.hpp
+++ b/ggml/src/ggml-sycl/dequantize.hpp
@@ -25,6 +25,28 @@ typedef void (*dequantize_kernel_f32_t)(const void * vx, const int64_t ib, const
 static inline void get_scale_min_k4(int j, const uint8_t * q, uint8_t & d, uint8_t & m);
 #endif
 
+static __dpct_inline__ void dequantize_q2_0(const void *vx, const int64_t ib,
+                                            const int iqs, dfloat2 &v) {
+    const block_q2_0 * x = (const block_q2_0 *) vx;
+
+    const dfloat d = x[ib].d;
+
+    const int byte_idx = iqs / 4;
+    const int shift = (iqs % 4) * 2;
+    const uint8_t vui = x[ib].qs[byte_idx];
+
+    v.x() = (vui >> shift) & 3;
+    v.y() = (vui >> (shift + 2)) & 3;
+
+#ifdef GGML_SYCL_F16
+    v.s0() = ((dfloat)v.s0() - 1.0f) * d;
+    v.s1() = ((dfloat)v.s1() - 1.0f) * d;
+#else
+    v.x() = ((dfloat)v.x() - 1.0f) * d;
+    v.y() = ((dfloat)v.y() - 1.0f) * d;
+#endif // GGML_SYCL_F16
+}
+
 static __dpct_inline__ void dequantize_q4_0(const void *vx, const int64_t ib,
                                             const int iqs, dfloat2 &v) {
     const block_q4_0 * x = (const block_q4_0 *) vx;
diff --git a/ggml/src/ggml-sycl/mmvq.cpp b/ggml/src/ggml-sycl/mmvq.cpp
index 7b1b3d467..863d34eab 100644
--- a/ggml/src/ggml-sycl/mmvq.cpp
+++ b/ggml/src/ggml-sycl/mmvq.cpp
@@ -1254,6 +1254,66 @@ static void mul_mat_vec_q1_0_q8_1_sycl_switch_ncols(
     }
 }
 
+static void mul_mat_vec_q2_0_q8_1_sycl(const void * vx, const void * vy,
+                                       float * dst, const int ncols,
+                                       const int nrows,
+                                       dpct::queue_ptr stream) {
+    GGML_ASSERT(ncols % QK2_0 == 0);
+    const int block_num_y = (nrows + GGML_SYCL_MMV_Y - 1) / GGML_SYCL_MMV_Y;
+    const sycl::range<3> block_nums(1, 1, block_num_y);
+    const sycl::range<3> block_dims(1, GGML_SYCL_MMV_Y, WARP_SIZE);
+
+    stream->submit([&](sycl::handler & cgh) {
+        cgh.parallel_for(
+            sycl::nd_range<3>(block_nums * block_dims, block_dims),
+            [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
+                mul_mat_vec_q<QK2_0, QI2_0, block_q2_0,
+                              VDR_Q2_0_Q8_1_MMVQ, vec_dot_q2_0_q8_1>(
+                    vx, vy, dst, ncols, nrows, item_ct1);
+            });
+    });
+}
+
+template <int ncols_dst>
+static void mul_mat_vec_q2_0_q8_1_sycl_ncols(
+        const void * vx, const void * vy, float * dst,
+        const int ncols, const int nrows,
+        const int stride_col_y, const int stride_col_dst,
+        dpct::queue_ptr stream) {
+    GGML_ASSERT(ncols % QK2_0 == 0);
+    const int block_num_y = (nrows + GGML_SYCL_MMV_Y - 1) / GGML_SYCL_MMV_Y;
+    const sycl::range<3> block_nums(1, 1, block_num_y);
+    const sycl::range<3> block_dims(1, GGML_SYCL_MMV_Y, WARP_SIZE);
+
+    stream->submit([&](sycl::handler & cgh) {
+        cgh.parallel_for(
+            sycl::nd_range<3>(block_nums * block_dims, block_dims),
+            [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
+                mul_mat_vec_q_ncols<QK2_0, QI2_0, block_q2_0,
+                                    VDR_Q2_0_Q8_1_MMVQ, vec_dot_q2_0_q8_1, ncols_dst>(
+                    vx, vy, dst, ncols, nrows, stride_col_y, stride_col_dst, item_ct1);
+            });
+    });
+}
+
+static void mul_mat_vec_q2_0_q8_1_sycl_switch_ncols(
+        const void * vx, const void * vy, float * dst,
+        const int ncols, const int nrows, const int ncols_dst,
+        const int stride_col_y, const int stride_col_dst,
+        dpct::queue_ptr stream) {
+    switch (ncols_dst) {
+        case 1: mul_mat_vec_q2_0_q8_1_sycl(vx, vy, dst, ncols, nrows, stream); break;
+        case 2: mul_mat_vec_q2_0_q8_1_sycl_ncols<2>(vx, vy, dst, ncols, nrows, stride_col_y, stride_col_dst, stream); break;
+        case 3: mul_mat_vec_q2_0_q8_1_sycl_ncols<3>(vx, vy, dst, ncols, nrows, stride_col_y, stride_col_dst, stream); break;
+        case 4: mul_mat_vec_q2_0_q8_1_sycl_ncols<4>(vx, vy, dst, ncols, nrows, stride_col_y, stride_col_dst, stream); break;
+        case 5: mul_mat_vec_q2_0_q8_1_sycl_ncols<5>(vx, vy, dst, ncols, nrows, stride_col_y, stride_col_dst, stream); break;
+        case 6: mul_mat_vec_q2_0_q8_1_sycl_ncols<6>(vx, vy, dst, ncols, nrows, stride_col_y, stride_col_dst, stream); break;
+        case 7: mul_mat_vec_q2_0_q8_1_sycl_ncols<7>(vx, vy, dst, ncols, nrows, stride_col_y, stride_col_dst, stream); break;
+        case 8: mul_mat_vec_q2_0_q8_1_sycl_ncols<8>(vx, vy, dst, ncols, nrows, stride_col_y, stride_col_dst, stream); break;
+        default: GGML_ABORT("unsupported ncols_dst=%d for Q2_0 multi-col MMVQ", ncols_dst);
+    }
+}
+
 static void mul_mat_vec_q2_K_q8_1_sycl(const void *vx, const void *vy,
                                        float *dst, const int ncols,
                                        const int nrows,
@@ -2194,6 +2254,20 @@ void ggml_sycl_op_mul_mat_vec_q(ggml_backend_sycl_context & ctx, const ggml_tens
                     mul_mat_vec_q1_0_q8_1_sycl(src0_dd_i, src1_ddq_i_bs, dst_dd_i_bs, ne00, row_diff, stream);
                 }
                 break;
+            case GGML_TYPE_Q2_0:
+                if (i == 0 && src1_ncols > 1 && src1_ncols <= 8) {
+                    const int stride_col_y   = src1_padded_col_size / QK8_1;
+                    const int stride_col_dst = dst->ne[0];
+                    GGML_SYCL_DEBUG("Calling mul_mat_vec_q2_0_q8_1_sycl_switch_ncols ncols=%d\n", (int)src1_ncols);
+                    mul_mat_vec_q2_0_q8_1_sycl_switch_ncols(
+                        src0_dd_i, src1_ddq_i, dst_dd_i, ne00, row_diff,
+                        src1_ncols, stride_col_y, stride_col_dst, stream);
+                    return;
+                } else if (i == 0 || src1_ncols == 1) {
+                    GGML_SYCL_DEBUG("Calling mul_mat_vec_q2_0_q8_1_sycl\n");
+                    mul_mat_vec_q2_0_q8_1_sycl(src0_dd_i, src1_ddq_i_bs, dst_dd_i_bs, ne00, row_diff, stream);
+                }
+                break;
             case GGML_TYPE_Q2_K:
                 if (i == 0 && src1_ncols > 1 && src1_ncols <= 8) {
                     const int stride_col_y   = src1_padded_col_size / QK8_1;
@@ -2503,6 +2577,11 @@ bool ggml_sycl_mul_mat_vec_q_id(
                 vx_base, vy, ids_dev, dst_base, ncols, nrows, n_experts_used,
                 expert_weight_stride, dst_row_stride, src1_row_stride, stream);
             return true;
+        case GGML_TYPE_Q2_0:
+            launch_mul_mat_vec_q_moe<QK2_0, QI2_0, block_q2_0, VDR_Q2_0_Q8_1_MMVQ, vec_dot_q2_0_q8_1>(
+                vx_base, vy, ids_dev, dst_base, ncols, nrows, n_experts_used,
+                expert_weight_stride, dst_row_stride, src1_row_stride, stream);
+            return true;
         case GGML_TYPE_Q2_K:
             launch_mul_mat_vec_q_moe<QK_K, QI2_K, block_q2_K, VDR_Q2_K_Q8_1_MMVQ, vec_dot_q2_K_q8_1>(
                 vx_base, vy, ids_dev, dst_base, ncols, nrows, n_experts_used,
diff --git a/ggml/src/ggml-sycl/vecdotq.hpp b/ggml/src/ggml-sycl/vecdotq.hpp
index 765fb7f15..c11a6e8f9 100644
--- a/ggml/src/ggml-sycl/vecdotq.hpp
+++ b/ggml/src/ggml-sycl/vecdotq.hpp
@@ -658,6 +658,40 @@ template <> struct reorder_vec_dot_q_sycl<GGML_TYPE_Q6_K> {
 #define VDR_Q4_0_Q8_1_MMVQ 2
 #define VDR_Q4_0_Q8_1_MMQ  4
 
+#define VDR_Q2_0_Q8_1_MMVQ 1
+
+template <int vdr>
+static __dpct_inline__ float vec_dot_q2_0_q8_1_impl(
+    const int * v,
+    const int * u,
+    const float & d2,
+    const sycl::half2 & ds8) {
+    int sumi = 0;
+
+#pragma unroll
+    for (int i = 0; i < vdr; ++i) {
+#pragma unroll
+        for (int j = 0; j < 4; ++j) {
+            const uint8_t q = (uint8_t) ((uint32_t) v[i] >> (8 * j));
+
+            // unpack 2-bit values to byte lanes (0..3), then apply zero-point
+            // correction with ds8f.y() below, mirroring the q4_0 style.
+            int vi = 0;
+            vi |= (((q >> 0) & 0x3) & 0xFF) << 0;
+            vi |= (((q >> 2) & 0x3) & 0xFF) << 8;
+            vi |= (((q >> 4) & 0x3) & 0xFF) << 16;
+            vi |= (((q >> 6) & 0x3) & 0xFF) << 24;
+
+            sumi = dpct::dp4a(vi, u[4 * i + j], sumi);
+        }
+    }
+
+    const sycl::float2 ds8f = ds8.convert<float, sycl::rounding_mode::automatic>();
+    // q2_0 has zero-point 1. Scale ds8f.y() by processed-lane ratio,
+    // consistent with q4_0's explicit zero-point subtraction style.
+    return d2 * (sumi * ds8f.x() - ((float) vdr / (float) QI2_0) * ds8f.y());
+}
+
 template <int vdr>
 static __dpct_inline__ float vec_dot_q4_0_q8_1_impl(const int * v, const int * u, const float & d4,
                                                     const sycl::half2 & ds8) {
@@ -882,6 +916,41 @@ vec_dot_q4_0_q8_1(const void *__restrict__ vbq,
     return vec_dot_q4_0_q8_1_impl<VDR_Q4_0_Q8_1_MMVQ>(v, u, bq4_0->d, bq8_1->ds);
 }
 
+static __dpct_inline__ float
+vec_dot_q2_0_q8_1(const void *__restrict__ vbq,
+                  const block_q8_1 *__restrict__ bq8_1, const int &iqs) {
+
+    const block_q2_0 * bq2_0 = (const block_q2_0 *) vbq;
+
+    int v[2 * VDR_Q2_0_Q8_1_MMVQ];
+    int u[8 * VDR_Q2_0_Q8_1_MMVQ];
+
+#pragma unroll
+    for (int i = 0; i < VDR_Q2_0_Q8_1_MMVQ; ++i) {
+        const int base = 4 * (iqs + i);
+
+        // Q2_0 has QK2_0 = 64 and uses 2 x QK8_1 blocks on the RHS.
+        v[2 * i + 0] = get_int_from_uint8(bq2_0->qs, iqs + i);
+        v[2 * i + 1] = get_int_from_uint8(bq2_0->qs, iqs + i + QI2_0);
+
+        u[8 * i + 0] = get_int_from_int8_aligned(bq8_1[0].qs, base + 0);
+        u[8 * i + 1] = get_int_from_int8_aligned(bq8_1[0].qs, base + 1);
+        u[8 * i + 2] = get_int_from_int8_aligned(bq8_1[0].qs, base + 2);
+        u[8 * i + 3] = get_int_from_int8_aligned(bq8_1[0].qs, base + 3);
+
+        u[8 * i + 4] = get_int_from_int8_aligned(bq8_1[1].qs, base + 0);
+        u[8 * i + 5] = get_int_from_int8_aligned(bq8_1[1].qs, base + 1);
+        u[8 * i + 6] = get_int_from_int8_aligned(bq8_1[1].qs, base + 2);
+        u[8 * i + 7] = get_int_from_int8_aligned(bq8_1[1].qs, base + 3);
+    }
+
+    const float sum0 = vec_dot_q2_0_q8_1_impl<VDR_Q2_0_Q8_1_MMVQ>(
+        v + 0, u + 0, bq2_0->d, bq8_1[0].ds);
+    const float sum1 = vec_dot_q2_0_q8_1_impl<VDR_Q2_0_Q8_1_MMVQ>(
+        v + VDR_Q2_0_Q8_1_MMVQ, u + 4 * VDR_Q2_0_Q8_1_MMVQ, bq2_0->d, bq8_1[1].ds);
+    return sum0 + sum1;
+}
+
 static __dpct_inline__ float
 vec_dot_q4_1_q8_1(const void *__restrict__ vbq,
                   const block_q8_1 *__restrict__ bq8_1, const int &iqs) {

From 1c5b89ff63151b5388cb5abb9357cebe446997c7 Mon Sep 17 00:00:00 2001
From: Neo Zhang <zhang.jianyu@outlook.com>
Date: Fri, 31 Jul 2026 14:20:28 +0800
Subject: [PATCH 24/30] sycl : support dev2dev memcpy by DEV2DEV_MEMCPY_FORWARD
 (#26234)

Co-authored-by: Neo Zhang Jianyu <jianyu.zhang@intel.com>
---
 docs/backend/SYCL.md             | 2 +-
 ggml/src/ggml-sycl/common.hpp    | 1 +
 ggml/src/ggml-sycl/ggml-sycl.cpp | 8 +++++++-
 3 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/docs/backend/SYCL.md b/docs/backend/SYCL.md
index c72dc3b3e..914d63af2 100644
--- a/docs/backend/SYCL.md
+++ b/docs/backend/SYCL.md
@@ -788,7 +788,7 @@ use 1 SYCL GPUs: [0] with Max compute units:512
 | Name              | Value            | Function                                                                                                                  |
 |-------------------|------------------|---------------------------------------------------------------------------------------------------------------------------|
 | GGML_SYCL_DEBUG   | 0 (default) or 1 | Enable log function by macro: GGML_SYCL_DEBUG                                                                             |
-| GGML_SYCL_DEV2DEV_MEMCPY | 0 (default) or 1 | Choose the SYCL or L0 API in dev2dev memory copy.<br>Value: <br>*  0: SYCL API (default)<br>* 1: L0 API -- L0 API is found to lead to abnormal crash in some case. This debug flag is used to check the issue.|
+| GGML_SYCL_DEV2DEV_MEMCPY | 0 (default), 1, 2 | Choose the method of dev2dev memory copy.<br>Value: <br>*  0: SYCL API (default), only support dGPUs.<br>* 1: L0 API -- Better performance, only support dGPUs, found to lead to abnormal crash in some case. <br>* 2: Host Forward -- Most stable method for all cases (including iGPU + dGPU*N), but with lower performance (-2% to -5%).<br>SYCL & L0 API are easy to be impacted by Intel GPU driver issue. When you meet the garbled output or crash issues in multiple GPUs case, try with this debug flag to work around or check the issue.|
 | GGML_SYCL_ENABLE_FLASH_ATTN | 1 (default) or 0| Enable Flash-Attention. It can reduce memory usage. The performance impact depends on the LLM.|
 | GGML_SYCL_ENABLE_OPT | 0 or 1 (default)| Enable optimize features for Intel GPUs. (Recommended to 0 for Intel devices older than Gen 10) |
 | GGML_SYCL_ENABLE_GRAPH | 0 (default) or 1 | Enable running computations through SYCL Graphs feature. Disabled by default because SYCL Graph is still on development, no better performance. |
diff --git a/ggml/src/ggml-sycl/common.hpp b/ggml/src/ggml-sycl/common.hpp
index f27ec5dd6..160331191 100644
--- a/ggml/src/ggml-sycl/common.hpp
+++ b/ggml/src/ggml-sycl/common.hpp
@@ -133,6 +133,7 @@ enum ggml_sycl_backend_gpu_mode {
 enum ggml_sycl_dev2dev_memcpy_mode {
   DEV2DEV_MEMCPY_SYCL = 0,
   DEV2DEV_MEMCPY_L0 = 1,
+  DEV2DEV_MEMCPY_FORWARD = 2
 };
 
 static_assert(sizeof(sycl::half) == sizeof(ggml_fp16_t), "wrong fp16 size");
diff --git a/ggml/src/ggml-sycl/ggml-sycl.cpp b/ggml/src/ggml-sycl/ggml-sycl.cpp
index 49bdc97d2..07ddfda88 100644
--- a/ggml/src/ggml-sycl/ggml-sycl.cpp
+++ b/ggml/src/ggml-sycl/ggml-sycl.cpp
@@ -274,6 +274,8 @@ static const char* dev2dev_int2str(int dev2dev) {
         return "SYCL API";
     } else if (dev2dev == DEV2DEV_MEMCPY_L0) {
         return "Level Zero API";
+    } else if (dev2dev == DEV2DEV_MEMCPY_FORWARD) {
+        return "Host Forward";
     } else {
         return "Unknown";
     }
@@ -684,7 +686,11 @@ static void dev2dev_memcpy(int device_dst, sycl::queue &q_dst, int device_src, s
     }
 
     // Host-staged copy
-    GGML_SYCL_DEBUG("[SYCL] dev2dev memcpy by host forward\n");
+    if(g_ggml_sycl_dev2dev_memcpy == DEV2DEV_MEMCPY_FORWARD) {
+        GGML_SYCL_DEBUG("[SYCL] dev2dev memcpy by host forward for setting GGML_SYCL_DEV2DEV_MEMCPY=2\n");
+    } else {
+        GGML_SYCL_DEBUG("[SYCL] dev2dev memcpy by host forward for SYCL/L0 fallback\n");
+    }
     char *host_buf = (char *)malloc(size);
     q_src.memcpy(host_buf, (const char *)ptr_src, size).wait();
     q_dst.memcpy((char *)ptr_dst, host_buf, size).wait();

From 1e22599522269c9d31f4ec3c4b0b75d01f07f9a7 Mon Sep 17 00:00:00 2001
From: Sachin Sharma <sachin@zettabolt.com>
Date: Fri, 31 Jul 2026 12:10:52 +0530
Subject: [PATCH 25/30] ggml-zendnn : group matmul direct API for mul_mat_id
 (#25918)

* ggml-zendnn : group matmul API for mul_mat_id

* ggml-zendnn : scale MUL_MAT_ID fallback threshold by expert count
---
 ggml/src/ggml-zendnn/CMakeLists.txt  |   2 +-
 ggml/src/ggml-zendnn/ggml-zendnn.cpp | 294 +++++++++++++++++++--------
 2 files changed, 215 insertions(+), 81 deletions(-)

diff --git a/ggml/src/ggml-zendnn/CMakeLists.txt b/ggml/src/ggml-zendnn/CMakeLists.txt
index e4ba9cfbd..87d721f6d 100644
--- a/ggml/src/ggml-zendnn/CMakeLists.txt
+++ b/ggml/src/ggml-zendnn/CMakeLists.txt
@@ -28,7 +28,7 @@ if (NOT ZENDNN_ROOT OR ZENDNN_ROOT STREQUAL "" OR ZENDNN_ROOT STREQUAL "OFF")
     ExternalProject_Add(
         zendnn
         GIT_REPOSITORY https://github.com/amd/ZenDNN.git
-        GIT_TAG 253b94ce0d7e9284c265fefb485714944caff9d3    # ZenDNN-2026-WW19
+        GIT_TAG 1f399a75cc0993778374a51bea49b64a57879595    # ZenDNN-2026-WW28
         PREFIX      ${ZENDNN_PREFIX}
         SOURCE_DIR  ${ZENDNN_SOURCE_DIR}
         BINARY_DIR  ${ZENDNN_BUILD_DIR}
diff --git a/ggml/src/ggml-zendnn/ggml-zendnn.cpp b/ggml/src/ggml-zendnn/ggml-zendnn.cpp
index 3c33dcb11..e6a9b51b7 100644
--- a/ggml/src/ggml-zendnn/ggml-zendnn.cpp
+++ b/ggml/src/ggml-zendnn/ggml-zendnn.cpp
@@ -30,6 +30,29 @@ zendnnl::common::data_type_t ggml_to_zendnn_type() {
     }
 }
 
+/**
+ * Builds the matmul_params shared by ggml_zendnn_matmul() and ggml_zendnn_group_matmul():
+ * dtype selection plus, for Q8_0 weights, dynamic-quant setup. Callers still need to set
+ * quant_params.src_scale.dims themselves, since that depends on the batch size(s) in use.
+ */
+template <typename TA, typename TB, typename TC>
+static zendnnl::lowoha::matmul::matmul_params ggml_zendnn_make_matmul_params(ggml_backend_zendnn_context * ctx) {
+    zendnnl::lowoha::matmul::matmul_params params;
+    params.dtypes.src = ggml_to_zendnn_type<TB>();
+    params.dtypes.wei = ggml_to_zendnn_type<TA>();
+    params.dtypes.dst = ggml_to_zendnn_type<TC>();
+    params.num_threads = ctx->n_threads;
+
+    if constexpr (std::is_same_v<TA, block_q8_0>) {
+        params.dtypes.compute = zendnnl::common::data_type_t::s8;
+        params.dynamic_quant = true;
+        params.quant_params.src_scale.buff = nullptr;
+        params.quant_params.src_scale.dt   = zendnnl::common::data_type_t::bf16;
+        params.packing.pack_format_b = 1;
+    }
+    return params;
+}
+
 /**
  * ZenDNN matmul: computes C = B * A.
  *
@@ -47,22 +70,12 @@ static bool ggml_zendnn_matmul(ggml_backend_zendnn_context * ctx, int64_t m, int
                                const TA * A, int64_t lda, const TB * B, int64_t ldb, TC * C,
                                int64_t ldc) {
 
-    zendnnl::lowoha::matmul::matmul_params params;
-    params.dtypes.src = ggml_to_zendnn_type<TB>();
-    params.dtypes.wei = ggml_to_zendnn_type<TA>();
-    params.dtypes.dst = ggml_to_zendnn_type<TC>();
-    params.num_threads = ctx->n_threads;
+    zendnnl::lowoha::matmul::matmul_params params = ggml_zendnn_make_matmul_params<TA, TB, TC>(ctx);
 
     zendnnl::lowoha::matmul::matmul_batch_params_t batch_params;
 
     if constexpr (std::is_same_v<TA, block_q8_0>) {
-        params.dtypes.compute = zendnnl::common::data_type_t::s8;
-        const int64_t num_groups = k / QK8_0;
-        params.dynamic_quant = true;
-        params.quant_params.src_scale.buff = nullptr;
-        params.quant_params.src_scale.dt   = zendnnl::common::data_type_t::bf16;
-        params.quant_params.src_scale.dims = {n, num_groups};
-        params.packing.pack_format_b = 1;
+        params.quant_params.src_scale.dims = {n, k / QK8_0};
     }
 
     zendnnl::error_handling::status_t status = zendnnl::lowoha::matmul::matmul_direct(
@@ -223,6 +236,99 @@ struct mmid_row_mapping {
     int32_t i2;
 };
 
+/**
+ * ZenDNN batched matmul: computes C[i] = B[i] * A[i] for every active expert i via a single
+ * group_matmul_direct() call. Batched analogue of ggml_zendnn_matmul() - see its docs for the
+ * per-expert A/B/C shape convention. m and k are shared by every expert; n (batch size) varies
+ * per expert, hence the vector.
+ */
+template <typename TA, typename TB, typename TC>
+static bool ggml_zendnn_group_matmul(ggml_backend_zendnn_context * ctx, int64_t m, int64_t k,
+                                     const std::vector<int64_t> & n,
+                                     const std::vector<const void *> & A, int64_t lda,
+                                     const std::vector<const void *> & B, int64_t ldb,
+                                     const std::vector<void *> & C, int64_t ldc) {
+
+    const int n_experts = n.size();
+
+    zendnnl::lowoha::matmul::matmul_params base_params = ggml_zendnn_make_matmul_params<TA, TB, TC>(ctx);
+
+    std::vector<char>         layout(n_experts, 'r');
+    std::vector<bool>         trans_a(n_experts, false);
+    std::vector<bool>         trans_b(n_experts, true);
+    std::vector<int>          batch_m(n_experts);
+    std::vector<int>          batch_n(n_experts, m);
+    std::vector<int>          batch_k(n_experts, k);
+    std::vector<float>        alpha(n_experts, 1.0f);
+    std::vector<float>        beta(n_experts, 0.0f);
+    std::vector<const void *> bias(n_experts, nullptr);
+    std::vector<int>          lda_v(n_experts, lda);
+    std::vector<int>          ldb_v(n_experts, ldb);
+    std::vector<int>          ldc_v(n_experts, ldc);
+    std::vector<bool>         is_wei_const(n_experts, true);
+    std::vector<zendnnl::lowoha::matmul::matmul_params> params(n_experts, base_params);
+
+    for (int i = 0; i < n_experts; i++) {
+        batch_m[i] = n[i];
+
+        // src_scale.dims depends on this expert's row count, unlike the rest of base_params
+        if constexpr (std::is_same_v<TA, block_q8_0>) {
+            params[i].quant_params.src_scale.dims = {n[i], k / QK8_0};
+        }
+    }
+
+    zendnnl::error_handling::status_t status = zendnnl::lowoha::matmul::group_matmul_direct(
+        layout, trans_a, trans_b, batch_m, batch_n, batch_k, alpha,
+        B, ldb_v, A, lda_v, bias, beta,
+        C, ldc_v, is_wei_const, params);
+
+    if (status != zendnnl::error_handling::status_t::success) {
+        GGML_LOG_ERROR("%s, ZenDNN group matmul failed: status=%d\n", __func__, static_cast<int>(status));
+        return false;
+    }
+    return true;
+}
+
+static bool ggml_zendnn_group_gemm(ggml_backend_zendnn_context * ctx, int64_t m, int64_t k,
+                                    const std::vector<int64_t> & n,
+                                    const std::vector<const void *> & A, int64_t lda,
+                                    const std::vector<const void *> & B, int64_t ldb,
+                                    const std::vector<void *> & C, int64_t ldc,
+                                    int Atype, int Btype, int Ctype) {
+
+    assert(m >= 0);
+    for (size_t i = 0; i < n.size(); i++) {
+        assert(n[i] >= 0);
+    }
+    assert(k >= 0);
+    assert(lda >= k);
+    assert(ldb >= k);
+    assert(ldc >= m);
+
+    // categorize types
+    switch (Atype) {
+        case GGML_TYPE_F32:
+            if (Btype != GGML_TYPE_F32 || Ctype != GGML_TYPE_F32)
+                return false;
+            return ggml_zendnn_group_matmul<float, float, float>(ctx, m, k, n, A, lda, B, ldb, C, ldc);
+        case GGML_TYPE_BF16:
+            if (Btype != GGML_TYPE_BF16)
+                return false;
+            if (Ctype == GGML_TYPE_BF16)
+                return ggml_zendnn_group_matmul<ggml_bf16_t, ggml_bf16_t, ggml_bf16_t>(
+                    ctx, m, k, n, A, lda, B, ldb, C, ldc);
+            if (Ctype == GGML_TYPE_F32)
+                return ggml_zendnn_group_matmul<ggml_bf16_t, ggml_bf16_t, float>(ctx, m, k, n, A, lda, B, ldb, C, ldc);
+            return false;
+        case GGML_TYPE_Q8_0:
+            if (Btype != GGML_TYPE_F32 || Ctype != GGML_TYPE_F32)
+                return false;
+            return ggml_zendnn_group_matmul<block_q8_0, float, float>(ctx, m, k, n, A, lda, B, ldb, C, ldc);
+        default:
+            return false; // unsupported type
+    }
+}
+
 static void ggml_zendnn_compute_forward_mul_mat_id(
     ggml_backend_zendnn_context * ctx,
     ggml_tensor * dst) {
@@ -262,7 +368,8 @@ static void ggml_zendnn_compute_forward_mul_mat_id(
     std::vector<int64_t> matrix_row_counts(n_as, 0);
     std::vector<std::vector<mmid_row_mapping>> matrix_rows(n_as);
 
-    int64_t max_rows = 0;
+    int64_t total_rows = 0;
+    int     n_active_experts = 0;
     // group rows by expert (preprocessing step)
     for (int64_t iid1 = 0; iid1 < ids->ne[1]; ++iid1) {
         for (int id = 0; id < n_ids; ++id) {
@@ -270,66 +377,74 @@ static void ggml_zendnn_compute_forward_mul_mat_id(
 
             GGML_ASSERT(i02 >= 0 && i02 < n_as);
 
+            if (matrix_row_counts[i02] == 0) {
+                n_active_experts++;
+            }
             matrix_rows[i02].push_back({id, iid1});
             matrix_row_counts[i02]++;
-            if (matrix_row_counts[i02] > max_rows) {
-                max_rows = matrix_row_counts[i02];
-            }
+            total_rows++;
         }
     }
 
-    if (max_rows == 0) {
+    if (total_rows == 0) {
         return; // no rows to process
     }
 
     const size_t row_size = ggml_row_size(vec_dot_type, ne10);
 
-    // size for converting src1 rows to vec_dot_type if needed
-    const size_t nbw1 = row_size;
-    const size_t nbw2 = nbw1 * ne11;
-    const size_t nbw3 = nbw2 * ne12;
-    const size_t src1_conv_size = (src1->type != vec_dot_type && src0->type != GGML_TYPE_Q8_0) ? ne13 * nbw3 : 0;
-
     // For Q8_0, src1 is always F32; the gather buffer must hold F32 rows (ne10*4 bytes),
     // not Q8_0-encoded rows (row_size ≈ ne10/32*34 bytes) — they differ by ~4x.
     const size_t f32_row_size = (size_t)ne10 * sizeof(float);
     const size_t gather_row_size = (src0->type == GGML_TYPE_Q8_0) ? f32_row_size : row_size;
 
+    if (src1->type != vec_dot_type && src0->type != GGML_TYPE_Q8_0) {
+        GGML_ASSERT(src1->type == GGML_TYPE_F32);
+    }
+
     // size for MoE gather/scatter buffers
-    const size_t wdata_cur_size = max_rows * gather_row_size;
-    const size_t dst_cur_size = max_rows * ggml_row_size(dst->type, ne01);
+    const size_t wdata_cur_size = total_rows * gather_row_size;
+    const size_t dst_cur_size = total_rows * ggml_row_size(dst->type, ne01);
 
     // allocate single buffer for all needs
-    const size_t total_size = src1_conv_size + wdata_cur_size + dst_cur_size;
+    const size_t total_size = wdata_cur_size + dst_cur_size;
     if (ctx->work_size < total_size) {
         ctx->work_data.reset(new char[total_size]);
         ctx->work_size = total_size;
     }
 
     // partition the buffer
-    char * work_data = ctx->work_data.get();
-    char * wdata_cur = work_data + src1_conv_size;
+    char * wdata_cur = ctx->work_data.get();
     char * dst_cur = wdata_cur + wdata_cur_size;
 
-    // ZenDNN requires FP32 for dynamic quantization, so conversion is skipped
-    if (src1->type != vec_dot_type && src0->type != GGML_TYPE_Q8_0) {
-        GGML_ASSERT(src1->type == GGML_TYPE_F32);
+    // per-expert data collected during gather, handed to ggml_zendnn_group_gemm() as one batch
+    std::vector<int64_t>      expert_row_count(n_active_experts);
+    std::vector<const void *> batch_src(n_active_experts);
+    std::vector<const void *> batch_wei(n_active_experts);
+    std::vector<void *>       batch_dst(n_active_experts);
 
-        #pragma omp parallel for collapse(3) num_threads(ctx->n_threads) schedule(static)
-        for (int64_t i13 = 0; i13 < ne13; ++i13) {
-            for (int64_t i12 = 0; i12 < ne12; ++i12) {
-                for (int64_t i11 = 0; i11 < ne11; ++i11) {
-                    const float * src1_f32 = (float *)((char *)src1->data + i11*nb11 + i12*nb12 + i13*nb13);
-                    void * src1_conv = (char *)work_data + i11*nbw1 + i12*nbw2 + i13*nbw3;
-                    from_float(src1_f32, src1_conv, ne10);
-                }
+    // precompute per-expert buffer offsets and batch indices for the parallel loop below
+    std::vector<int64_t> expert_wdata_off(n_as, 0);
+    std::vector<int64_t> expert_dst_off(n_as, 0);
+    std::vector<int>     expert_batch_idx(n_as, -1);
+    {
+        int64_t w_off = 0;
+        int64_t d_off = 0;
+        int     batch_idx = 0;
+        for (int64_t cur_a = 0; cur_a < n_as; ++cur_a) {
+            if (matrix_row_counts[cur_a] == 0) {
+                continue;
             }
+            expert_wdata_off[cur_a] = w_off;
+            expert_dst_off[cur_a]   = d_off;
+            expert_batch_idx[cur_a] = batch_idx;
+            w_off += matrix_row_counts[cur_a] * gather_row_size;
+            d_off += matrix_row_counts[cur_a] * ggml_row_size(dst->type, ne01);
+            batch_idx++;
         }
     }
 
-    const void * wdata = (src1->type == vec_dot_type || src0->type == GGML_TYPE_Q8_0) ? src1->data : work_data;
-
-    // process each expert with gather -> gemm -> scatter pattern
+    // gather + inline-convert input rows into each expert's batch slot
+    #pragma omp parallel for num_threads(ctx->n_threads) schedule(static)
     for (int64_t cur_a = 0; cur_a < n_as; ++cur_a) {
         const int64_t cne1 = matrix_row_counts[cur_a];
 
@@ -337,42 +452,57 @@ static void ggml_zendnn_compute_forward_mul_mat_id(
             continue;
         }
 
-        const char * src0_cur = (const char *) src0->data + cur_a*nb02;
+        const int64_t w_off     = expert_wdata_off[cur_a];
+        const int64_t d_off     = expert_dst_off[cur_a];
+        const int     batch_idx = expert_batch_idx[cur_a];
 
-        // gather input rows for this expert
-        #pragma omp parallel for num_threads(ctx->n_threads) schedule(static)
         for (int64_t ir1 = 0; ir1 < cne1; ++ir1) {
             const mmid_row_mapping & row_mapping = matrix_rows[cur_a][ir1];
-            const int64_t id = row_mapping.i1;
+            const int64_t id  = row_mapping.i1;
             const int64_t i11 = id % ne11;
             const int64_t i12 = row_mapping.i2;
 
-            std::memcpy(
-                wdata_cur + ir1 * gather_row_size,
-                (const char *) wdata + (i11 + i12*ne11) * gather_row_size,
-                gather_row_size
-            );
+            const char * src_row = (const char *) src1->data + i11*nb11 + i12*nb12;
+            void *       dst_row = wdata_cur + w_off + ir1 * gather_row_size;
+
+            if (src1->type != vec_dot_type && src0->type != GGML_TYPE_Q8_0) {
+                from_float((const float *) src_row, dst_row, ne10);
+            } else {
+                // no conversion: src1 already matches vec_dot_type, or src0 is Q8_0, whose
+                // ZenDNN dynamic quantization requires the row to stay in F32
+                std::memcpy(dst_row, src_row, gather_row_size);
+            }
         }
 
-        // batched gemm for all tokens in this expert
-        if (!ggml_zendnn_gemm(ctx,
-                              ne01,       // m
-                              cne1,       // n
-                              ne10,       // k
-                              src0_cur,
-                              ne00,       // lda
-                              wdata_cur,
-                              ne10,       // ldb
-                              dst_cur,
-                              ne01,       // ldc
-                              src0->type,
-                              src0->type == GGML_TYPE_Q8_0 ? GGML_TYPE_F32 : vec_dot_type,
-                              dst->type)) {
-            GGML_ABORT("%s: ZenDNN gemm failed\n", __func__);
+        expert_row_count[batch_idx] = cne1;
+        batch_src[batch_idx]        = wdata_cur + w_off;
+        batch_wei[batch_idx]        = (const char *) src0->data + cur_a * nb02;
+        batch_dst[batch_idx]        = dst_cur + d_off;
+    }
+
+    if (!ggml_zendnn_group_gemm(ctx,
+                                ne01,             // m
+                                ne10,             // k
+                                expert_row_count, // n (per expert)
+                                batch_wei, ne00,  // A: weights, lda
+                                batch_src, ne10,  // B: input, ldb
+                                batch_dst, ne01,  // C: output, ldc
+                                src0->type,
+                                src0->type == GGML_TYPE_Q8_0 ? GGML_TYPE_F32 : vec_dot_type,
+                                dst->type))
+        GGML_ABORT("%s: ZenDNN group gemm failed\n", __func__);
+
+    // scatter output rows to destination
+    #pragma omp parallel for num_threads(ctx->n_threads) schedule(static)
+    for (int64_t cur_a = 0; cur_a < n_as; ++cur_a) {
+        const int64_t cne1 = matrix_row_counts[cur_a];
+
+        if (cne1 == 0) {
+            continue;
         }
 
-        // scatter output rows to destination
-        #pragma omp parallel for num_threads(ctx->n_threads) schedule(static)
+        const int64_t d_off = expert_dst_off[cur_a];
+
         for (int64_t ir1 = 0; ir1 < cne1; ++ir1) {
             const mmid_row_mapping & row_mapping = matrix_rows[cur_a][ir1];
             const int64_t id = row_mapping.i1;
@@ -381,7 +511,7 @@ static void ggml_zendnn_compute_forward_mul_mat_id(
 
             std::memcpy(
                 (char *) dst->data + i1*nb1 + i2*nb2,
-                dst_cur + ir1 * ggml_row_size(dst->type, ne01),
+                dst_cur + d_off + ir1 * ggml_row_size(dst->type, ne01),
                 ggml_row_size(dst->type, ne01)
             );
         }
@@ -591,22 +721,26 @@ static bool ggml_backend_zendnn_device_supports_op(ggml_backend_dev_t dev, const
                 if(K <= 256 || N <= 128 || M <= 96) {
                     return false;
                 }
+
+                // MUL_MAT_ID's gather+matmul+scatter approach favors a moderate expert count
+                if (op->op == GGML_OP_MUL_MAT_ID) {
+                    const int64_t n_experts = weights->ne[2];
+                    const int64_t max_experts = 32;
+                    if (n_experts > max_experts) {
+                        return false;
+                    }
+
+                    // fall back once the average rows per expert (N / n_experts) is too thin
+                    // to amortize each per-expert GEMM's overhead
+                    if (N / n_experts <= 32) {
+                        return false;
+                    }
+                }
             }
             else if (ne0 < min_batch || ne1 < min_batch || ne10 < min_batch) {
                 return false;
             }
 
-            // MUL_MAT_ID performs best with a moderate number of experts due to its
-            // gather + batched matmul + scatter approach. Future versions will leverage
-            // ZenDNN's grouped_gemm for better scalability with larger expert counts:
-            // https://github.com/amd/ZenDNN/blob/main/docs/operator/lowoha_group_gemm_operator.md
-            if (op->op == GGML_OP_MUL_MAT_ID) {
-                const int64_t n_experts = weights->ne[2];
-                const int64_t max_experts = 32;
-                if (n_experts > max_experts) {
-                    return false;
-                }
-            }
             switch (weights->type) {
                 case GGML_TYPE_F32:
                 case GGML_TYPE_BF16:

From 69e62fc77c911da169cc8726b490028d53bb90fe Mon Sep 17 00:00:00 2001
From: fairydreaming <166155368+fairydreaming@users.noreply.github.com>
Date: Fri, 31 Jul 2026 09:03:30 +0200
Subject: [PATCH 26/30] llama : enforce the same K and V cache types for
 DeepSeek V4; enable FA if V cache is quantized (#25871)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

* llama : enforce the same K and V cache types for DeepSeek V4; enable FA if V cache is quantized

* llama : enforce the same K and V cache types for MLA models

---------

Co-authored-by: Stanisław Szymczyk <sszymczy@gmail.com>
---
 src/llama-context.cpp | 21 ++++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)

diff --git a/src/llama-context.cpp b/src/llama-context.cpp
index 3e0af3f31..5ef7becf6 100644
--- a/src/llama-context.cpp
+++ b/src/llama-context.cpp
@@ -3553,6 +3553,22 @@ llama_context * llama_init_from_model(
         }
     }
 
+    if ((model->hparams.is_mla() || model->arch == LLM_ARCH_DEEPSEEK4) && params.type_k != params.type_v) {
+        LLAMA_LOG_ERROR("%s: model does not support different K (%s) and V (%s) cache types\n", __func__, ggml_type_name(params.type_k), ggml_type_name(params.type_v));
+        return nullptr;
+    }
+
+    if (ggml_is_quantized(params.type_v) && params.flash_attn_type != LLAMA_FLASH_ATTN_TYPE_ENABLED) {
+        if (params.flash_attn_type == LLAMA_FLASH_ATTN_TYPE_AUTO) {
+            LLAMA_LOG_INFO("%s: enabling flash_attn since it is required for quantized V cache\n", __func__);
+            params.flash_attn_type = LLAMA_FLASH_ATTN_TYPE_ENABLED;
+        }
+        if (params.flash_attn_type == LLAMA_FLASH_ATTN_TYPE_DISABLED) {
+            LLAMA_LOG_ERROR("%s: quantized V cache requires flash_attn to be enabled\n", __func__);
+            return nullptr;
+        }
+    }
+
     if (params.flash_attn_type != LLAMA_FLASH_ATTN_TYPE_DISABLED && ggml_is_quantized(params.type_k)) {
         const uint32_t blck_size = ggml_blck_size(params.type_k);
         for (uint32_t il = 0; il < model->hparams.n_layer(); ++il) {
@@ -3575,11 +3591,6 @@ llama_context * llama_init_from_model(
         }
     }
 
-    if (ggml_is_quantized(params.type_v) && params.flash_attn_type == LLAMA_FLASH_ATTN_TYPE_DISABLED) {
-        LLAMA_LOG_ERROR("%s: V cache quantization requires flash_attn\n", __func__);
-        return nullptr;
-    }
-
     if (params.pooling_type != LLAMA_POOLING_TYPE_UNSPECIFIED &&
         params.pooling_type != model->hparams.pooling_type) {
         //user-specified pooling-type is different from the model default

From d5d3e05bf8d2a29b71270b1af9d538dc2545b02a Mon Sep 17 00:00:00 2001
From: Neo Zhang <zhang.jianyu@outlook.com>
Date: Fri, 31 Jul 2026 15:25:16 +0800
Subject: [PATCH 27/30] [SYCL] support the missed types in cpy (#26005)

* support the missed types in cpy

* use correct funct

* rm unused code
---
 ggml/src/ggml-sycl/cpy.cpp | 183 ++++++++++++++++++++++++++-----------
 ggml/src/ggml-sycl/cpy.hpp |  33 +++++++
 2 files changed, 164 insertions(+), 52 deletions(-)

diff --git a/ggml/src/ggml-sycl/cpy.cpp b/ggml/src/ggml-sycl/cpy.cpp
index 5d0f9a89f..55e076172 100644
--- a/ggml/src/ggml-sycl/cpy.cpp
+++ b/ggml/src/ggml-sycl/cpy.cpp
@@ -8,7 +8,6 @@
 #include "ggml-sycl/presets.hpp"
 #include "ggml.h"
 
-
 static void cpy_1_f32_f32(const char * cxi, char * cdsti) {
     const float * xi   = (const float *) cxi;
     float *       dsti = (float *) cdsti;
@@ -151,6 +150,20 @@ static void cpy_blck_q8_0_f32(const char * cxi, char * cdsti) {
     }
 }
 
+static void cpy_blck_q2_0_f32(const char * cxi, char * cdsti) {
+    const block_q2_0 * xi = (const block_q2_0 *) cxi;
+    float * cdstf = (float *) cdsti;
+
+    const float d = xi->d;
+
+    for (int j = 0; j < QK2_0; ++j) {
+        const int byte_index = j / 4;
+        const int bit_offset = (j % 4) * 2;
+        const int q = (xi->qs[byte_index] >> bit_offset) & 0x3;
+        cdstf[j] = (float) (q - 1) * d;
+    }
+}
+
 
 
 template <dequantize_kernel_t dequant, int qk> static void cpy_blck_q_f32(const char * cxi, char * cdsti) {
@@ -256,7 +269,7 @@ static void ggml_cpy_f16_f32_sycl(const char * cx, char * cdst, const int ne, co
         stream->parallel_for(
             sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                               sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-            [=](sycl::nd_item<3> item_ct1) {
+            [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                 cpy_f32_f16<cpy_1_f16_f32>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
             });
@@ -274,7 +287,7 @@ static void ggml_cpy_f32_f32_sycl(const char * cx, char * cdst, const int ne, co
         stream->parallel_for(
             sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                               sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-            [=](sycl::nd_item<3> item_ct1) {
+            [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                 cpy_f32_f16<cpy_1_f32_f32>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
             });
@@ -292,7 +305,7 @@ static void ggml_cpy_f32_f16_sycl(const char * cx, char * cdst, const int ne, co
         stream->parallel_for(
             sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                               sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-            [=](sycl::nd_item<3> item_ct1) {
+            [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                 cpy_f32_f16<cpy_1_f32_f16>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
             });
@@ -308,7 +321,7 @@ static void ggml_cpy_f32_i32_sycl(const char * cx, char * cdst, const int ne, co
         stream->parallel_for(
             sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                               sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-            [=](sycl::nd_item<3> item_ct1) {
+            [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                 cpy_f32_f16<cpy_1_f32_i32>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
             });
@@ -324,7 +337,7 @@ static void ggml_cpy_i32_f32_sycl(const char * cx, char * cdst, const int ne, co
         stream->parallel_for(
             sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                               sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-            [=](sycl::nd_item<3> item_ct1) {
+            [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                 cpy_f32_f16<cpy_1_i32_f32>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
             });
@@ -338,7 +351,7 @@ static void ggml_cpy_f32_q8_0_sycl(const char * cx, char * cdst, const int ne, c
     GGML_ASSERT(ne % QK8_0 == 0);
     const int num_blocks = ne / QK8_0;
     stream->parallel_for(sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
-                         [=](sycl::nd_item<3> item_ct1) {
+                         [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                              cpy_f32_q<cpy_blck_f32_q8_0, QK8_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03,
                                                                  ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
                          });
@@ -350,12 +363,25 @@ static void ggml_cpy_q8_0_f32_sycl(const char * cx, char * cdst, const int ne, c
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ne;
     stream->parallel_for(sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
-                         [=](sycl::nd_item<3> item_ct1) {
+                         [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                              cpy_q_f32<cpy_blck_q8_0_f32, QK8_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03,
                                                                  ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
                          });
 }
 
+static void ggml_cpy_q2_0_f32_sycl(const char * cx, char * cdst, const int ne, const int ne00, const int ne01,
+                                   const int ne02, const int nb00, const int nb01, const int nb02, const int nb03,
+                                   const int ne10, const int ne11, const int ne12, const int nb10, const int nb11,
+                                   const int nb12, const int nb13, queue_ptr stream) {
+    const int num_blocks = ne;
+    stream->parallel_for(
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
+            cpy_q_f32<cpy_blck_q2_0_f32, QK2_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11,
+                                                ne12, nb10, nb11, nb12, nb13, item_ct1);
+        });
+}
+
 static void ggml_cpy_f32_q4_0_sycl(const char * cx, char * cdst, const int ne, const int ne00, const int ne01,
                                    const int ne02, const int nb00, const int nb01, const int nb02, const int nb03,
                                    const int ne10, const int ne11, const int ne12, const int nb10, const int nb11,
@@ -363,7 +389,7 @@ static void ggml_cpy_f32_q4_0_sycl(const char * cx, char * cdst, const int ne, c
     GGML_ASSERT(ne % QK4_0 == 0);
     const int num_blocks = ne / QK4_0;
     stream->parallel_for(sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
-                         [=](sycl::nd_item<3> item_ct1) {
+                         [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                              cpy_f32_q<cpy_blck_f32_q4_0, QK4_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03,
                                                                  ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
                          });
@@ -375,7 +401,8 @@ static void ggml_cpy_q4_0_f32_sycl(const char * cx, char * cdst, const int ne, c
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ne;
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_f32<cpy_blck_q_f32<dequantize_q4_0, QK4_0>, QK4_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02,
                                                                      nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13,
                                                                      item_ct1);
@@ -389,7 +416,7 @@ static void ggml_cpy_f32_q4_1_sycl(const char * cx, char * cdst, const int ne, c
     GGML_ASSERT(ne % QK4_1 == 0);
     const int num_blocks = ne / QK4_1;
     stream->parallel_for(sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
-                         [=](sycl::nd_item<3> item_ct1) {
+                         [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                              cpy_f32_q<cpy_blck_f32_q4_1, QK4_1>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03,
                                                                  ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
                          });
@@ -401,7 +428,8 @@ static void ggml_cpy_q4_1_f32_sycl(const char * cx, char * cdst, const int ne, c
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ne;
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_f32<cpy_blck_q_f32<dequantize_q4_1, QK4_1>, QK4_1>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02,
                                                                      nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13,
                                                                      item_ct1);
@@ -415,7 +443,7 @@ static void ggml_cpy_f32_q5_0_sycl(const char * cx, char * cdst, const int ne, c
     GGML_ASSERT(ne % QK5_0 == 0);
     const int num_blocks = ne / QK5_0;
     stream->parallel_for(sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
-                         [=](sycl::nd_item<3> item_ct1) {
+                         [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
                              cpy_f32_q<cpy_blck_f32_q5_0, QK5_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03,
                                                                  ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
                          });
@@ -427,7 +455,8 @@ static void ggml_cpy_q5_0_f32_sycl(const char * cx, char * cdst, const int ne, c
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ne;
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_f32<cpy_blck_q_f32<dequantize_q5_0, QK5_0>, QK5_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02,
                                                                      nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13,
                                                                      item_ct1);
@@ -441,7 +470,7 @@ static void ggml_cpy_f32_q5_1_sycl(const char * cx, char * cdst, const int ne, c
     GGML_ASSERT(ne % QK5_1 == 0);
     const int num_blocks = ne / QK5_1;
     stream->parallel_for(sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
-                         [=](sycl::nd_item<3> item_ct1) {
+                         [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                              cpy_f32_q<cpy_blck_f32_q5_1, QK5_1>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03,
                                                                  ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
                          });
@@ -453,7 +482,8 @@ static void ggml_cpy_q5_1_f32_sycl(const char * cx, char * cdst, const int ne, c
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ne;
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_f32<cpy_blck_q_f32<dequantize_q5_1, QK5_1>, QK5_1>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02,
                                                                      nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13,
                                                                      item_ct1);
@@ -466,7 +496,8 @@ static void ggml_cpy_mxfp4_f32_sycl(const char * cx, char * cdst, const int ne,
                                     const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ne;
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
             cpy_q_f32<cpy_blck_q_f32<dequantize_mxfp4, QK_MXFP4>, QK_MXFP4>(cx, cdst, ne, ne00, ne01, ne02, nb00,
                                                                              nb01, nb02, nb03, ne10, ne11, ne12,
                                                                              nb10, nb11, nb12, nb13, item_ct1);
@@ -480,7 +511,8 @@ static void ggml_cpy_f32_iq4_nl_sycl(const char * cx, char * cdst, const int ne,
     GGML_ASSERT(ne % QK4_NL == 0);
     const int num_blocks = ne / QK4_NL;
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
             cpy_f32_q<cpy_blck_f32_iq4_nl, QK4_NL>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11,
                                                    ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
@@ -526,7 +558,7 @@ static void ggml_cpy_f16_q4_0_sycl(const char * cx, char * cdst, const int ne, c
     GGML_ASSERT(ne % QK4_0 == 0);
     const int num_blocks = ne / QK4_0;
     stream->parallel_for(sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
-                         [=](sycl::nd_item<3> item_ct1) {
+                         [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                              cpy_f32_q<cpy_blck_f16_q4_0, QK4_0>(cx, cdst, ne, ne00, ne01, ne02,
                                                                  nb00, nb01, nb02, nb03,
                                                                  ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
@@ -540,7 +572,7 @@ static void ggml_cpy_f16_q4_1_sycl(const char * cx, char * cdst, const int ne, c
     GGML_ASSERT(ne % QK4_1 == 0);
     const int num_blocks = ne / QK4_1;
     stream->parallel_for(sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
-                         [=](sycl::nd_item<3> item_ct1) {
+                         [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                              cpy_f32_q<cpy_blck_f16_q4_1, QK4_1>(cx, cdst, ne, ne00, ne01, ne02,
                                                                  nb00, nb01, nb02, nb03,
                                                                  ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
@@ -554,7 +586,7 @@ static void ggml_cpy_f16_q5_0_sycl(const char * cx, char * cdst, const int ne, c
     GGML_ASSERT(ne % QK5_0 == 0);
     const int num_blocks = ne / QK5_0;
     stream->parallel_for(sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks), sycl::range<3>(1, 1, 1)),
-                         [=](sycl::nd_item<3> item_ct1) {
+                         [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                              cpy_f32_q<cpy_blck_f16_q5_0, QK5_0>(cx, cdst, ne, ne00, ne01, ne02,
                                                                  nb00, nb01, nb02, nb03,
                                                                  ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
@@ -564,6 +596,7 @@ static void ggml_cpy_f16_q5_0_sycl(const char * cx, char * cdst, const int ne, c
 static bool ggml_sycl_is_quantized_type(enum ggml_type type) {
     switch (type) {
         case GGML_TYPE_Q1_0:
+        case GGML_TYPE_Q2_0:
         case GGML_TYPE_Q4_0:
         case GGML_TYPE_Q4_1:
         case GGML_TYPE_Q5_0:
@@ -594,6 +627,7 @@ static bool ggml_sycl_is_quantized_type(enum ggml_type type) {
 static bool ggml_sycl_can_quantize_rows_sycl(enum ggml_type type) {
     switch (type) {
         case GGML_TYPE_Q1_0:
+        case GGML_TYPE_Q2_0:
         case GGML_TYPE_Q4_0:
         case GGML_TYPE_Q4_1:
         case GGML_TYPE_Q5_0:
@@ -651,7 +685,8 @@ static void ggml_sycl_quantize_rows_q(const char * cx, char * cdst, const int64_
     constexpr int block_size = 256;
     const int64_t grid_size = ceil_div(total_blocks, (int64_t) block_size);
 
-    stream->parallel_for(sycl::nd_range<1>(grid_size * block_size, block_size), [=](sycl::nd_item<1> item_ct1) {
+    stream->parallel_for(sycl::nd_range<1>(grid_size * block_size, block_size),
+                        [=](sycl::nd_item<1> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
         const int64_t block_idx = item_ct1.get_global_linear_id();
         if (block_idx >= total_blocks) {
             return;
@@ -708,6 +743,11 @@ static void ggml_sycl_quantize_rows_sycl(const char * cx, char * cdst, const ggm
                                                                             nb02, nb03, ne10, ne11, ne12, nb10, nb11,
                                                                             nb12, nb13, stream);
             break;
+        case GGML_TYPE_Q2_0:
+            ggml_sycl_quantize_rows_q<SrcScalar, cpy_blck_f32_q2_0, QK2_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01,
+                                                                            nb02, nb03, ne10, ne11, ne12, nb10, nb11,
+                                                                            nb12, nb13, stream);
+            break;
         case GGML_TYPE_Q5_1:
             ggml_sycl_quantize_rows_q<SrcScalar, cpy_blck_f32_q5_1, QK5_1>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01,
                                                                             nb02, nb03, ne10, ne11, ne12, nb10, nb11,
@@ -760,7 +800,7 @@ static void ggml_cpy_f16_f16_sycl(const char * cx, char * cdst, const int ne, co
         stream->parallel_for(
             sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                               sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-            [=](sycl::nd_item<3> item_ct1) {
+            [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                 cpy_f32_f16<cpy_1_f16_f16>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
             });
@@ -779,7 +819,7 @@ static void ggml_cpy_i16_i16_sycl(const char * cx, char * cdst, const int ne, co
         stream->parallel_for(
             sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                               sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-            [=](sycl::nd_item<3> item_ct1) {
+            [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                 cpy_f32_f16<cpy_1_i16_i16>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
             });
@@ -798,7 +838,7 @@ static void ggml_cpy_i32_i32_sycl(const char * cx, char * cdst, const int ne, co
         stream->parallel_for(
             sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                               sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-            [=](sycl::nd_item<3> item_ct1) {
+            [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
                 cpy_f32_f16<cpy_1_i32_i32>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
             });
@@ -812,7 +852,8 @@ static void ggml_cpy_q8_0_q8_0(const char * cx, char * cdst, const int ne, const
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
         sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
-                              sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+                              sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+                              [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_q8_0, QK8_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -825,7 +866,8 @@ static void ggml_cpy_q5_0_q5_0(const char * cx, char * cdst, const int ne, const
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
         sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
-                              sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+                              sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+                              [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_q5_0, QK5_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -839,7 +881,8 @@ static void ggml_cpy_q5_1_q5_1(const char * cx, char * cdst, const int ne, const
 
     stream->parallel_for(
         sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
-                              sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+                              sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+                              [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_q5_1, QK5_1>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -851,7 +894,8 @@ static void ggml_cpy_q4_0_q4_0(const char * cx, char * cdst, const int ne, const
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_q4_0, QK4_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -864,7 +908,8 @@ static void ggml_cpy_q4_1_q4_1(const char * cx, char * cdst, const int ne, const
 
    const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
    stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_q4_1, QK4_1>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -875,18 +920,32 @@ static void ggml_cpy_q1_0_q1_0(const char * cx, char * cdst, const int ne, const
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
             cpy_q_q<block_q1_0, QK1_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
 
+static void ggml_cpy_q2_0_q2_0(const char * cx, char * cdst, const int ne, const int ne00, const int ne01,
+                                   const int ne02, const int nb00, const int nb01, const int nb02, const int nb03,
+                                   const int ne10, const int ne11, const int ne12, const int nb10, const int nb11,
+                                   const int nb12, const int nb13, queue_ptr stream) {
+    const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
+    stream->parallel_for(
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
+            cpy_q_q<block_q2_0, QK2_0>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
+        });
+}
+
 static void ggml_cpy_mxfp4_mxfp4(const char * cx, char * cdst, const int ne, const int ne00, const int ne01,
                                    const int ne02, const int nb00, const int nb01, const int nb02, const int nb03,
                                    const int ne10, const int ne11, const int ne12, const int nb10, const int nb11,
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
             cpy_q_q<block_mxfp4, QK_MXFP4>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -897,7 +956,8 @@ static void ggml_cpy_nvfp4_nvfp4(const char * cx, char * cdst, const int ne, con
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_nvfp4, QK_NVFP4>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -908,7 +968,8 @@ static void ggml_cpy_q2_K_q2_K(const char * cx, char * cdst, const int ne, const
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_q2_K, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -919,7 +980,8 @@ static void ggml_cpy_q3_K_q3_K(const char * cx, char * cdst, const int ne, const
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_q3_K, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -930,7 +992,8 @@ static void ggml_cpy_q4_K_q4_K(const char * cx, char * cdst, const int ne, const
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_q4_K, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -941,7 +1004,8 @@ static void ggml_cpy_q5_K_q5_K(const char * cx, char * cdst, const int ne, const
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_q5_K, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -952,7 +1016,8 @@ static void ggml_cpy_q6_K_q6_K(const char * cx, char * cdst, const int ne, const
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_q6_K, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -963,7 +1028,8 @@ static void ggml_cpy_iq2_xxs_iq2_xxs(const char * cx, char * cdst, const int ne,
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_iq2_xxs, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -974,7 +1040,8 @@ static void ggml_cpy_iq2_xs_iq2_xs(const char * cx, char * cdst, const int ne, c
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_iq2_xs, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -985,7 +1052,8 @@ static void ggml_cpy_iq2_s_iq2_s(const char * cx, char * cdst, const int ne, con
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_iq2_s, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -996,7 +1064,8 @@ static void ggml_cpy_iq3_xxs_iq3_xxs(const char * cx, char * cdst, const int ne,
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_iq3_xxs, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -1007,7 +1076,8 @@ static void ggml_cpy_iq1_s_iq1_s(const char * cx, char * cdst, const int ne, con
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_iq1_s, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -1018,7 +1088,8 @@ static void ggml_cpy_iq1_m_iq1_m(const char * cx, char * cdst, const int ne, con
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_iq1_m, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -1029,7 +1100,8 @@ static void ggml_cpy_iq4_nl_iq4_nl(const char * cx, char * cdst, const int ne, c
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_iq4_nl, QK4_NL>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -1040,7 +1112,8 @@ static void ggml_cpy_iq3_s_iq3_s(const char * cx, char * cdst, const int ne, con
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_iq3_s, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -1051,7 +1124,8 @@ static void ggml_cpy_iq4_xs_iq4_xs(const char * cx, char * cdst, const int ne, c
                                    const int nb12, const int nb13, queue_ptr stream) {
     const int num_blocks = ceil_div(ne, SYCL_CPY_BLOCK_SIZE);
     stream->parallel_for(
-        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)), [=](sycl::nd_item<3> item_ct1) {
+        sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE), sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_q_q<block_iq4_xs, QK_K>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, item_ct1);
         });
 }
@@ -1065,7 +1139,7 @@ static void ggml_cpy_f32_bf16_sycl(const char * cx, char * cdst, const int ne, c
     stream->parallel_for(
         sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                           sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-        [=](sycl::nd_item<3> item_ct1) {
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_f32_f16<cpy_1_f32_bf16>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
         });
@@ -1079,7 +1153,7 @@ static void ggml_cpy_bf16_f32_sycl(const char * cx, char * cdst, const int ne, c
     stream->parallel_for(
         sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                           sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-        [=](sycl::nd_item<3> item_ct1) {
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_f32_f16<cpy_1_bf16_f32>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
         });
@@ -1093,7 +1167,7 @@ static void ggml_cpy_bf16_bf16_sycl(const char * cx, char * cdst, const int ne,
     stream->parallel_for(
         sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                           sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-        [=](sycl::nd_item<3> item_ct1) {
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_f32_f16<cpy_1_bf16_bf16>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                             nb10, nb11, nb12, nb13, item_ct1);
         });
@@ -1107,7 +1181,7 @@ static void ggml_cpy_f16_bf16_sycl(const char * cx, char * cdst, const int ne, c
     stream->parallel_for(
         sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                           sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-        [=](sycl::nd_item<3> item_ct1) {
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_f32_f16<cpy_1_f16_bf16>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
         });
@@ -1121,7 +1195,7 @@ static void ggml_cpy_bf16_f16_sycl(const char * cx, char * cdst, const int ne, c
     stream->parallel_for(
         sycl::nd_range<3>(sycl::range<3>(1, 1, num_blocks) * sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE),
                           sycl::range<3>(1, 1, SYCL_CPY_BLOCK_SIZE)),
-        [=](sycl::nd_item<3> item_ct1) {
+        [=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]]{
             cpy_f32_f16<cpy_1_bf16_f16>(cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
                                            nb10, nb11, nb12, nb13, item_ct1);
         });
@@ -1213,6 +1287,9 @@ void ggml_sycl_cpy(ggml_backend_sycl_context & ctx, const ggml_tensor * src0, co
     } else if (src0->type == GGML_TYPE_Q8_0 && src1->type == GGML_TYPE_F32) {
         ggml_cpy_q8_0_f32_sycl(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10,
                                nb11, nb12, nb13, main_stream);
+    } else if (src0->type == GGML_TYPE_Q2_0 && src1->type == GGML_TYPE_F32) {
+        ggml_cpy_q2_0_f32_sycl(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12,
+                               nb10, nb11, nb12, nb13, main_stream);
     } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_Q5_0) {
         ggml_cpy_f32_q5_0_sycl(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10,
                                nb11, nb12, nb13, main_stream);
@@ -1243,6 +1320,8 @@ void ggml_sycl_cpy(ggml_backend_sycl_context & ctx, const ggml_tensor * src0, co
         ggml_cpy_q4_1_q4_1(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
     } else if (src0->type == GGML_TYPE_Q1_0 && src1->type == GGML_TYPE_Q1_0) {
         ggml_cpy_q1_0_q1_0(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
+    } else if (src0->type == GGML_TYPE_Q2_0 && src1->type == GGML_TYPE_Q2_0) {
+        ggml_cpy_q2_0_q2_0(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
     } else if (src0->type == GGML_TYPE_MXFP4 && src1->type == GGML_TYPE_MXFP4) {
         ggml_cpy_mxfp4_mxfp4(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
     } else if (src0->type == GGML_TYPE_NVFP4 && src1->type == GGML_TYPE_NVFP4) {
diff --git a/ggml/src/ggml-sycl/cpy.hpp b/ggml/src/ggml-sycl/cpy.hpp
index c4cfd961d..34bae1b2d 100644
--- a/ggml/src/ggml-sycl/cpy.hpp
+++ b/ggml/src/ggml-sycl/cpy.hpp
@@ -70,6 +70,39 @@ inline void cpy_blck_f32_q1_0(const char * cxi, char * cdsti) {
     }
 }
 
+inline int round_nearest_int(float x) {
+    return (int)(x >= 0.0f ? x + 0.5f : x - 0.5f);
+}
+
+inline void cpy_blck_f32_q2_0(const char * cxi, char * cdsti) {
+    const float * xi   = (const float *) cxi;
+    block_q2_0 *  dsti = (block_q2_0 *) cdsti;
+
+    float amax = 0.0f;
+
+    for (int j = 0; j < QK2_0; ++j) {
+        amax = sycl::fmax(amax, sycl::fabs((float) xi[j]));
+    }
+
+    const float d  = amax;
+    const float id = d > 0.0f ? 1.0f / d : 0.0f;
+
+    dsti->d = d;
+
+    for (int j = 0; j < QK2_0 / 4; ++j) {
+        dsti->qs[j] = 0;
+    }
+
+    for (int j = 0; j < QK2_0; ++j) {
+        int q = round_nearest_int(xi[j] * id) + 1;
+        q = dpct::max(0, dpct::min(3, q));
+
+        const int byte_index = j / 4;
+        const int bit_offset = (j % 4) * 2;
+        dsti->qs[byte_index] |= (uint8_t) q << bit_offset;
+    }
+}
+
 inline int best_index_mxfp4(const float x, const float e) {
     int best_index = 0;
     float best_err = sycl::fabs((float) (kvalues_mxfp4[0] * e - x));

From 9d9a6d29f6b981cc7f41983d26e56485c6af1811 Mon Sep 17 00:00:00 2001
From: Ozymandias_EBON <112784549+johnkarlhill@users.noreply.github.com>
Date: Fri, 31 Jul 2026 02:43:16 -0500
Subject: [PATCH 28/30] =?UTF-8?q?SYCL:=20add=20oneMKL=20GEMM=20flash=20att?=
 =?UTF-8?q?ention=20for=20XMX-accelerated=20prompt=20proc=E2=80=A6=20(#250?=
 =?UTF-8?q?25)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

* SYCL: add oneMKL GEMM flash attention for XMX-accelerated prompt processing

* fattn-mkl: fix interleaved dst layout in normalize kernel

- Fix mkl_fa_normalize_head: use interleaved dst layout
  ((query * n_q_heads + head) * DV) matching TILE's
  flash_attn_combine_results. Previously used dense head-major
  layout which wrote head outputs to wrong addresses, corrupting
  attention for all models except Qwen3.6-27B (where GQA=6 heads
  were sparse enough to avoid visible overlap).

- Remove 7 redundant stream->wait() calls — SYCL in-order queue
  already serializes pure SYCL kernel dependencies. Retain only
  the 4 MKL GEMM ↔ SYCL handshake barriers (oneMKL GEMM uses its
  own internal queue that does not respect SYCL in-order).

- Remove unused dst_row_stride, diagnostic clutter, and dead
  K/V hex dump (fa_diag block in fattn-mkl.cpp).

- Add MKL_FA_DISABLE=1 env var for A/B testing.
- Add FA-DISP watchdog (MKL_FA_DEBUG=1) and FA-DIAG output
  fingerprint (MKL_FA_DIAG=1) in fattn.cpp.

Tested: Gemma-4-26B, Gemma-4-31B, Qwen3.6-27B, Qwen3.6-35B-A3B
Perf (B70/Battlemage, 32K, q8_0 KV):
  Gemma-4-26B:  1473 t/s MKL vs 746 TILE (1.97x)
  Qwen3.6-27B:   609 t/s MKL vs 330 TILE (1.85x)

Co-Authored-By: Claude Code on DeepSeek-v4-Pro

* Thank you for the review feedback: rename env vars, use GGML_LOG_INFO, document in SYCL.md

Completed the following:
- Rename MKL_FA_DISABLE → GGML_SYCL_ENABLE_MKL_FA (inverted: 0 to disable)
- Rename MKL_FA_DEBUG → GGML_SYCL_MKL_FA_DEBUG
- Rename MKL_FA_DIAG → GGML_SYCL_MKL_FA_DIAG
- Replace fprintf(stderr, ...) / fflush(stderr) with GGML_LOG_INFO() macro
- Document all three env vars in docs/backend/SYCL.md under Runtime
- Add comment explaining MKL FA activation trigger (flash-attn + quantized
  KV cache + batch-size >= 1024 + n_kv >= 1024)

Resolves review feedback from arthw.
Again, thank you!!!

Co-Authored-By: Claude Code on DeepSeek-v4-Pro

* Thank you for the review feedback round 2: use ggml_sycl_get_env, remove dup waits, gate perf macros

- Replace raw getenv() with ggml_sycl_get_env() in all 4 env-var checks
  (fattn.cpp: GGML_SYCL_ENABLE_MKL_FA, GGML_SYCL_MKL_FA_DEBUG,
   GGML_SYCL_MKL_FA_DIAG; fattn-mkl.cpp: GGML_SYCL_MKL_FA_DEBUG)
- Remove duplicated stream->wait() before ev.wait_and_throw() in GEMM
  KQ and GEMM VKQ — ev.wait_and_throw() already waits for completion
- Gate MKL_ACCUM macro behind do_print so timing accumulators are
  no-ops in normal operation
- Remove redundant MIT/Intel copyright header from fattn-mkl.cpp
- Remove unused #include <cfloat>
- Expand SYCL.md MKL FA docs with step-by-step activation trigger
  and example llama-cli command

Again, thank you!!!

Co-Authored-By: Claude Code on DeepSeek-v4-Pro

* fattn-mkl: enable MKL FA for all KV cache types

Remove the quantized-only restriction on MKL activation — the MKL
kernel converts any non-F16 K/V to F16 via to_fp16_sycl before GEMM,
so F16 (default), BF16, and F32 caches all benefit from XMX hardware
acceleration.  The type restriction was an unnecessary gate.

Before (F16/BF16 default cache + FA on at 32K prefill): ~356 t/s (TILE path)
After:  ~670 t/s (MKL path, matching quantized-cache baseline)

Minimal change: two conditions removed, one comment updated in fattn.cpp.
No kernel or conversion code changes — the dequant pipeline already
covers all types.

* fattn-mkl: rename mkl_disable -> mkl_enable for clarity

* fattn-mkl: refine MKL FA dispatch gates

Three changes:
1. Remove quantized-only restriction - MKL FA activates for all
   KV cache types (F16 default, BF16, F32, quantized).  The MKL
   kernel converts non-F16 K/V via to_fp16_sycl before GEMM.
2. Rename mkl_disable -> mkl_enable to match env var
   (GGML_SYCL_ENABLE_MKL_FA).
3. Replace batch-size threshold with Q->ne[1] >= 32 gate.
   Keeps TG (Q=1) and MTP drafts (Q=3-8) on VEC path where
   fused kernel beats MKL launch overhead.  Routes all
   multi-token prefill through XMX-accelerated GEMM.

Production data confirms Q patterns: 1-8 TG, 32-127 cache reuse,
128+ full reprocess.  At 32K F16/BF16 FA-on: 356 -> 670 t/s.

* ggml-sycl: fix F16 cache + MKL FA multi-turn corruption; add gate guards

Two changes:

1. Always copy F16 K/V to dense row-major buffers before MKL GEMM.
   Previously F16 was read in-place with raw tensor strides. During
   multi-turn conversations, the accumulated KV cache had different
   stride properties than a fresh prefill, producing corrupted outputs.
   Now dense F16 gets a fast memcpy; interleaved (Gemma) gets a strided
   copy kernel. This matches what the quantized paths already did through
   to_fp16_sycl.

2. Gate MKL FA on unsupported op params (max_bias, logit_softcap, batch
   dim mismatch) and pathological F16 strides (nb[1] not a multiple of
   ne[0]*2). These conditions would previously crash inside the MKL
   kernel. Pathological strides (test-only) and ALiBi/softcap fall
   through to TILE/VEC which handle them correctly.

The stride check uses modulo rather than equality, so both dense
(nb1 == ne0*2) and interleaved (nb1 == H * ne0*2) pass — all real
models use these layouts. Only test cases with overlapping rows
(nb1=32 or nb1=75 for ne0=40) are blocked.

Thanks to hmscider for the oneDNN FA PR (#25222) which surfaced the
same insight: always normalize inputs to contiguous F16 before GEMM.

Co-Authored-By: Claude Code using DeepSeek-V4-Pro <noreply@anthropic.com>

* fattn-mkl: fix quant+GQA KV strides, tighten MKL gate, add K>=1024 tests

Adding K>=1024 flash-attn test cases surfaced several MKL bugs:

- Quant K/V with a padded seq-view (real KV cache) used the wrong
  strides in the dequant path... only the true Gemma interleave
  layout should reconstruct strides. nb[2] vs ne[1]*nb[1]
- Gate was firing on shapes the kernel doesn't handle: head_dim < 64
  or not a multiple of 64, MHA, attention sinks, and
  bf16 decode... fell through to vec which no bf16 case.

Gate MKL to the validated envelope: gqa>=2, head_dim 64 through 512
(has to be a multiple of 64) with matching K/V head size, mask,
no sinks/alibi/softcap... everything else falls back to tile.
Covers Qwen Dense/MoE and Gemma4 Dense/MoE

Ran test-backend-ops -o FLASH_ATTN_EXT: 3641/3641 pass.
Perplexity unchanged... 6.7267 MKL vs 6.7290 stock using
Qwen 27b q5_k_xl

* Update ggml/src/ggml-sycl/fattn.cpp

Co-authored-by: Neo Zhang <zhang.jianyu@outlook.com>

* Update ggml/src/ggml-sycl/fattn.cpp

Co-authored-by: Neo Zhang <zhang.jianyu@outlook.com>

* Update ggml/src/ggml-sycl/fattn.cpp

Co-authored-by: Neo Zhang <zhang.jianyu@outlook.com>

* fattn-mkl: bound attention scratch so it doesn't grow with batch or context... also dropped the bf16 comment in fattn.cpp per arthw review.

* Update ggml/src/ggml-sycl/fattn-mkl.cpp

Co-authored-by: Neo Zhang <zhang.jianyu@outlook.com>

* Update ggml/src/ggml-sycl/fattn-mkl.cpp

Co-authored-by: Neo Zhang <zhang.jianyu@outlook.com>

* apply arthw suggestions: enum for dequant modes, macro for wg_size, env-var one-liners

---------

Co-authored-by: Claude Code using DeepSeek-V4-Pro <noreply@anthropic.com>
Co-authored-by: Neo Zhang <zhang.jianyu@outlook.com>
---
 docs/backend/SYCL.md             |   3 +
 ggml/src/ggml-sycl/fattn-mkl.cpp | 690 +++++++++++++++++++++++++++++++
 ggml/src/ggml-sycl/fattn.cpp     | 121 ++++++
 ggml/src/ggml-sycl/fattn.hpp     |   2 +
 tests/test-backend-ops.cpp       |  12 +
 5 files changed, 828 insertions(+)
 create mode 100644 ggml/src/ggml-sycl/fattn-mkl.cpp

diff --git a/docs/backend/SYCL.md b/docs/backend/SYCL.md
index 914d63af2..814e541e1 100644
--- a/docs/backend/SYCL.md
+++ b/docs/backend/SYCL.md
@@ -797,6 +797,9 @@ use 1 SYCL GPUs: [0] with Max compute units:512
 | GGML_SYCL_FA_ONEDNN | 1 (default) or 0 | Enable the oneDNN fused SDPA (flash-attention) path on supported GPUs. Set to 0 to always use the native SYCL flash-attention kernel. |
 | GGML_SYCL_FA_ONEDNN_MAX_KV | 0 (default, disabled) or positive integer | By default (0), all sequences are handled by the oneDNN fused SDPA path, regardless of KV length; a positive value caps that length, past which sequences fall back to the native kernel. If GPU driver watchdog resets (DEVICE_LOST) occur during long-context inference, set this near the context depth where they start, e.g. 24576. |
 | GGML_SYCL_ENABLE_VMM | 0 or 1 (default) | Enable the virtual-memory device pool. |
+| GGML_SYCL_ENABLE_MKL_FA | 1 (default) or 0 | Enable oneMKL GEMM flash attention for XMX-accelerated prompt processing with quantized KV cache. Automatically activates during prefill (prompt processing) when all conditions are met: (1) flash-attn enabled (`-fa` or `--flash-attn on`), (2) KV cache quantized (`--cache-type-k q8_0 --cache-type-v q8_0` or other `*_0/*_1` types), (3) batch size ≥ 1024 (`--batch-size 1024`), (4) prompt length ≥ 1024 tokens. Set to 0 to force the TILE kernel for A/B testing. Example minimum command: `llama-cli -m model.gguf -fa -ngl 99 --cache-type-k q8_0 --cache-type-v q8_0 --batch-size 1024 -p "your prompt"` |
+| GGML_SYCL_MKL_FA_DEBUG | 0 (default) or 1 | Enable per-call diagnostic logging for MKL flash attention: GEMM/softmax timings, interleaved-head detection, and buffer memory usage. |
+| GGML_SYCL_MKL_FA_DIAG | 0 (default) or 1 | Enable output fingerprinting for MKL flash attention. Dumps the first 64 float output values for the first 6 FA calls with n_kv ≥ 1024, labeled with kernel type (MKL/TILE/VEC) for cross-kernel comparison. |
 | GGML_SYCL_ENABLE_FUSION | 0 or 1 (default) | Enable fused-kernel dispatch in graph compute (currently top-k MoE gating). |
 | ZES_ENABLE_SYSMAN | 0 (default) or 1 | Support to get free memory of GPU by sycl::aspect::ext_intel_free_memory.<br>Recommended to use when --split-mode = layer |
 | UR_L0_ENABLE_RELAXED_ALLOCATION_LIMITS | 0 (default) or 1 | Allow SYCL/Unified Runtime Level Zero device allocations larger than 4 GiB. llama.cpp's direct Level Zero allocation path requests the relaxed maximum-size limit itself when GGML_SYCL_ENABLE_LEVEL_ZERO=1. |
diff --git a/ggml/src/ggml-sycl/fattn-mkl.cpp b/ggml/src/ggml-sycl/fattn-mkl.cpp
new file mode 100644
index 000000000..fc22b7bdb
--- /dev/null
+++ b/ggml/src/ggml-sycl/fattn-mkl.cpp
@@ -0,0 +1,690 @@
+// Flash attention via oneMKL GEMM (XMX-accelerated).
+// Uses column_major::gemm for Q*K^T and S*V matmuls
+// with an online softmax SYCL kernel.
+//
+// All GQA query heads sharing a KV head are batched into single
+// GEMM calls, amortizing MKL launch overhead across K and V reuse.
+//
+
+#include "common.hpp"
+#include "fattn-common.hpp"
+#include "fattn-buffers.hpp"
+#include "convert.hpp"
+#include "fattn.hpp"
+
+#include <oneapi/mkl.hpp>
+#include <cstdio>
+#include <chrono>
+
+#define MKL_FA_CHUNK_SIZE_KV 8192
+
+// Number of query rows processed per tile. The score buffers (KQ_f32, S_f16)
+// are sized q_tile_rows * chunk_size, so this bounds their footprint
+// regardless of batch size (n_query_rows = n_queries * gqa_ratio). A typical
+// single-ubatch prefill (e.g. ubatch 1024 * gqa 8 = 8192 rows) is exactly one
+// tile, so it runs with no extra iterations. Larger batches tile and stay
+// bounded. Override with GGML_SYCL_MKL_FA_Q_TILE.
+#define MKL_FA_Q_TILE 8192
+
+#define MKL_FA_WG_SIZE 256
+
+using oneapi::mkl::transpose;
+using oneapi::mkl::blas::column_major::gemm;
+
+// ---------------------------------------------------------------------------
+// Helpers
+// ---------------------------------------------------------------------------
+
+// Pack all GQA Q heads for one KV head into fp16, applying q_scale.
+// Launches one kernel per GQA group — each kernel copies exactly
+// n_queries * DKQ elements using the per-group dst offset and
+// per-head source stride.
+static void mkl_fa_pack_q_fp16(
+    dpct::queue_ptr stream,
+    sycl::half * __restrict dst,
+    const float * __restrict q_src,
+    int n_queries, int n_query_rows, int DKQ,
+    int gqa_ratio, int kvh_base_head,
+    float q_scale, int64_t q_row_stride, int64_t q_head_stride,
+    int64_t wg_size) {
+
+    for (int iqg = 0; iqg < gqa_ratio; iqg++) {
+        int     iqh       = kvh_base_head + iqg;
+        sycl::half * dst_g = dst + (int64_t)iqg * n_queries * DKQ;
+
+        const int64_t n_elem = (int64_t)n_queries * DKQ;
+        const int64_t wg = ((n_elem + wg_size - 1) / wg_size) * wg_size;
+
+        stream->submit([&](sycl::handler & cgh) {
+            cgh.parallel_for(sycl::nd_range<1>(wg, wg_size),
+                [=](sycl::nd_item<1> item) {
+                    int64_t e = item.get_global_id(0);
+                    if (e >= n_elem) return;
+
+                    int64_t q = e / DKQ;
+                    int64_t d = e - q * DKQ;
+
+                    // Stride-aware source offset: handles permuted,
+                    // sliced, or contiguous Q tensor layouts.
+                    int64_t src_off = d
+                        + q * q_row_stride
+                        + (int64_t)iqh * q_head_stride;
+
+                    dst_g[e] = sycl::half(
+                        q_src[src_off] * q_scale);
+                });
+        });
+    }
+}
+
+// Zero-initialize the online softmax state arrays.
+// KQ_max → -inf, KQ_sum → 0, VKQ_accum → 0.
+// Merged into one kernel to avoid per-array launch overhead.
+static void mkl_fa_init_softmax_state(
+    dpct::queue_ptr stream,
+    float * kmax, float * ksum, float * vacc,
+    int n_query_rows, int DV, int64_t wg_size) {
+
+    const float    neg_inf   = -1e30f;
+    const int64_t  n_maxsum  = n_query_rows;
+    const int64_t  n_vacc    = (int64_t)n_query_rows * DV;
+    const int64_t  total     = (n_vacc > n_maxsum) ? n_vacc : n_maxsum;
+    const int64_t  wg = ((total + wg_size - 1) / wg_size) * wg_size;
+
+    stream->submit([&](sycl::handler & cgh) {
+        cgh.parallel_for(sycl::nd_range<1>(wg, wg_size),
+            [=](sycl::nd_item<1> item) {
+                int64_t i = item.get_global_id(0);
+                if (i < n_maxsum) {
+                    kmax[i] = neg_inf;
+                    ksum[i] = 0.0f;
+                }
+                if (i < n_vacc) {
+                    vacc[i] = 0.0f;
+                }
+            });
+    });
+}
+
+// Online softmax over one KV chunk for a tile of GQA query rows.
+// The tile spans absolute rows [q0, q0 + q_rows). Score buffers
+// (KQ_f32/S_f16) are indexed RELATIVE to the tile; the persistent state
+// (VKQ_accum/KQ_max/KQ_sum) and mask are indexed by ABSOLUTE row.
+// For each row: find local max → rescale previous VKQ_accum →
+// compute exp(s - max) → write S_f16 → update running max/sum.
+static void mkl_fa_online_softmax_chunk(
+    dpct::queue_ptr stream,
+    float * __restrict KQ_f32,
+    sycl::half * __restrict S_f16,
+    float * __restrict KQ_max,
+    float * __restrict KQ_sum,
+    float * __restrict VKQ_accum,
+    int q0, int q_rows, int n_queries, int DV,
+    int chunk_size, int chunk_start,
+    int kvh_head, int gqa_ratio,
+    const sycl::half * mask_data, int64_t mask_head_stride,
+    int64_t mask_row_stride, int mask_n_heads,
+    float logit_softcap, int64_t wg_size) {
+
+    const int64_t wg = ((q_rows + wg_size - 1) / wg_size) * wg_size;
+
+    stream->submit([&](sycl::handler & cgh) {
+        cgh.parallel_for(sycl::nd_range<1>(wg, wg_size),
+            [=](sycl::nd_item<1> item) {
+                int jc_rel = item.get_global_id(0);
+                if (jc_rel >= q_rows) return;
+                int jc_abs = q0 + jc_rel;
+
+                const int gqa_group = jc_abs / n_queries;
+                const int q_row     = jc_abs % n_queries;
+
+                // Score buffers are tile-local (relative index).
+                const float * __restrict KQ_row = KQ_f32
+                    + jc_rel * (int64_t)chunk_size;
+                // Persistent accumulator is full-sized (absolute index).
+                float * __restrict vkq = VKQ_accum
+                    + jc_abs * (int64_t)DV;
+
+                const sycl::half * mask_h = nullptr;
+                int64_t m_stride = 0;
+                if (mask_data) {
+                    int m_head = (mask_n_heads > 1)
+                        ? (kvh_head + gqa_group) : 0;
+                    mask_h   = mask_data + (int64_t)m_head * mask_head_stride;
+                    m_stride = mask_row_stride;
+                }
+
+                // Row-wise local maximum (softcap before mask)
+                float local_max = -1e30f;
+                for (int i = 0; i < chunk_size; i++) {
+                    float s = KQ_row[i];
+                    if (logit_softcap != 0.0f) {
+                        s = logit_softcap * sycl::tanh(s);
+                    }
+                    if (mask_h) {
+                        s += (float)mask_h[q_row * m_stride
+                            + (chunk_start + i)];
+                    }
+                    if (s > local_max) local_max = s;
+                }
+
+                // Rescale previous accumulator by exp(old_max - new_max)
+                float old_max = KQ_max[jc_abs];
+                float new_max = (old_max > local_max) ? old_max : local_max;
+                float rescale = (old_max < -1e29f) ? 1.0f
+                    : sycl::native::exp(old_max - new_max);
+
+                for (int v = 0; v < DV; v++) {
+                    vkq[v] *= rescale;
+                }
+
+                // Softmax and write S_f16 (tile-local index)
+                float local_sum = 0.0f;
+                sycl::half * __restrict S_row = S_f16
+                    + jc_rel * (int64_t)chunk_size;
+
+                for (int i = 0; i < chunk_size; i++) {
+                    float s = KQ_row[i];
+                    if (logit_softcap != 0.0f) {
+                        s = logit_softcap * sycl::tanh(s);
+                    }
+                    if (mask_h) {
+                        s += (float)mask_h[q_row * m_stride
+                            + (chunk_start + i)];
+                    }
+                    float val = sycl::native::exp(s - new_max);
+                    S_row[i] = sycl::half(val);
+                    local_sum += val;
+                }
+
+                KQ_sum[jc_abs] = KQ_sum[jc_abs] * rescale + local_sum;
+                KQ_max[jc_abs] = new_max;
+            });
+    });
+}
+
+// Write one GQA group's normalized output to its destination head.
+static void mkl_fa_normalize_head(
+    dpct::queue_ptr stream,
+    float * __restrict dst_batch,
+    const float * __restrict VKQ_accum,
+    const float * __restrict KQ_sum,
+    int iqh, int n_queries, int DV, int n_q_heads,
+    int64_t src_offset, int64_t wg_size) {
+
+    const int64_t wg = ((n_queries + wg_size - 1) / wg_size) * wg_size;
+
+    stream->submit([&](sycl::handler & cgh) {
+        cgh.parallel_for(sycl::nd_range<1>(wg, wg_size),
+            [=](sycl::nd_item<1> item) {
+                int jc = item.get_global_id(0);
+                if (jc >= n_queries) return;
+
+                int     ksum_idx = (int)(src_offset / DV) + jc;
+                float   inv_sum  = 1.0f / KQ_sum[ksum_idx];
+                const float * __restrict src = VKQ_accum
+                    + src_offset + jc * (int64_t)DV;
+                // Interleaved dst layout (matching TILE):
+                // rows alternate between heads, then increment query.
+                // offset = (query * n_q_heads + head) * DV
+                float * __restrict dst_row = dst_batch
+                    + ((int64_t)jc * n_q_heads + iqh) * (int64_t)DV;
+
+                for (int v = 0; v < DV; v++) {
+                    dst_row[v] = src[v] * inv_sum;
+                }
+            });
+    });
+}
+
+// ---------------------------------------------------------------------------
+// Per-chunk dequant
+//
+// Rather than dequantizing all of K/V up front (footprint scales with
+// context), we dequant one KV-head chunk at a time into a dense
+// [this_chunk x D] fp16 buffer (row-major, lda = D). The source address of
+// element (head=ikvh, row=chunk_start+r, col=c) decomposes into independent
+// linear terms head_off(ikvh) + row_off(chunk_start) + (r,c), so slicing a
+// chunk is a clean pointer offset in every layout case. The true-Gemma-
+// interleave vs padded-seq-view distinction is resolved once when the
+// descriptor is built; slicing does not reintroduce it.
+// ---------------------------------------------------------------------------
+enum mkl_fa_kv_desc_mode {
+    MKL_FA_KV_MODE_F16_DENSE       = 0,
+    MKL_FA_KV_MODE_F16_INTERLEAVED = 1,
+    MKL_FA_KV_MODE_QUANT_CONTIG    = 2,
+    MKL_FA_KV_MODE_QUANT_NC        = 3,
+};
+
+struct mkl_fa_kv_desc {
+    const char *         data = nullptr;
+    ggml_type            type = GGML_TYPE_F16;
+    int64_t              D    = 0;      // ne[0]
+    int64_t              nb1  = 0;      // byte stride, seq dim
+    int64_t              nb2  = 0;      // byte stride, head dim
+    mkl_fa_kv_desc_mode  mode = MKL_FA_KV_MODE_F16_DENSE;
+    int64_t              ts   = 0;      // type size (mode 3 base offset)
+    int64_t              s01  = 0;      // nc row stride in blocks (mode 3)
+    int64_t              s02  = 0;      // nc head stride in blocks (mode 3)
+};
+
+static mkl_fa_kv_desc mkl_fa_make_desc(const ggml_tensor * T, bool interleaved, int n_kv_heads) {
+    mkl_fa_kv_desc d;
+    d.data = (const char *)T->data;
+    d.type = T->type;
+    d.D    = T->ne[0];
+    d.nb1  = (int64_t)T->nb[1];
+    d.nb2  = (int64_t)T->nb[2];
+    d.ts   = (int64_t)ggml_type_size(T->type);
+
+    if (T->type == GGML_TYPE_F16) {
+        d.mode = interleaved ? MKL_FA_KV_MODE_F16_INTERLEAVED
+                             : MKL_FA_KV_MODE_F16_DENSE;
+    } else if (ggml_is_contiguously_allocated(T) && !interleaved) {
+        d.mode = MKL_FA_KV_MODE_QUANT_CONTIG;
+    } else {
+        d.mode = MKL_FA_KV_MODE_QUANT_NC;
+        const int64_t bs          = (int64_t)ggml_blck_size(T->type);
+        const int64_t blk_per_row = T->ne[0] / bs;
+        // True Gemma interleave packs heads within a row (nb[2] < ne[1]*nb[1])
+        // → reconstruct physical strides. Padded seq-views (nb[2] > ne[1]*nb[1])
+        // already have correct physical strides.
+        const bool gemma = interleaved &&
+            ((int64_t)T->nb[2] < (int64_t)T->ne[1] * (int64_t)T->nb[1]);
+        if (gemma) {
+            d.s01 = (int64_t)n_kv_heads * blk_per_row;
+            d.s02 = blk_per_row;
+        } else {
+            d.s01 = d.nb1 / d.ts;
+            d.s02 = d.nb2 / d.ts;
+        }
+    }
+    return d;
+}
+
+// Dequant one KV-head chunk into a dense [this_chunk x D] fp16 buffer.
+static void mkl_fa_dequant_chunk(
+    dpct::queue_ptr stream, const mkl_fa_kv_desc & d, ggml_tensor * dst_ctx,
+    sycl::half * out, int ikvh, int chunk_start, int this_chunk) {
+
+    const int64_t D = d.D;
+    switch (d.mode) {
+        case MKL_FA_KV_MODE_F16_DENSE: {
+            const char * base = d.data + (int64_t)ikvh * d.nb2
+                + (int64_t)chunk_start * d.nb1;
+            stream->memcpy(out, base, (size_t)this_chunk * D * sizeof(sycl::half));
+            break;
+        }
+        case MKL_FA_KV_MODE_F16_INTERLEAVED: {
+            const char * base = d.data + (int64_t)ikvh * d.nb2
+                + (int64_t)chunk_start * d.nb1;
+            const int64_t row_halfs = d.nb1 / (int64_t)sizeof(sycl::half);
+            const sycl::half * src = (const sycl::half *)base;
+            stream->parallel_for(
+                sycl::range<2>((size_t)this_chunk, (size_t)D),
+                [=](sycl::item<2> it) {
+                    int64_t r = it.get_id(0);
+                    int64_t c = it.get_id(1);
+                    out[r * D + c] = src[r * row_halfs + c];
+                });
+            break;
+        }
+        case MKL_FA_KV_MODE_QUANT_CONTIG: {
+            const char * base = d.data + (int64_t)ikvh * d.nb2
+                + (int64_t)chunk_start * d.nb1;
+            to_fp16_sycl_t to_fp16 = ggml_get_to_fp16_sycl(d.type, dst_ctx);
+            to_fp16(base, out, (int64_t)this_chunk * D, stream);
+            break;
+        }
+        default: {  // MKL_FA_KV_MODE_QUANT_NC
+            to_fp16_nc_sycl_t to_fp16 = ggml_get_to_fp16_nc_sycl(d.type);
+            const int64_t base_blocks = (int64_t)ikvh * d.s02
+                + (int64_t)chunk_start * d.s01;
+            const char * base = d.data + base_blocks * d.ts;
+            // ne02 = ne03 = 1 → s02/s03 inert; head+chunk offset carried by base.
+            to_fp16(base, out, D, this_chunk, 1, 1, d.s01, d.s02, d.s02, stream);
+            break;
+        }
+    }
+}
+
+// ---------------------------------------------------------------------------
+// MKL Flash Attention orchestrator
+//
+// Pipeline: dequantize K/V → for each KV head:
+//   pack GQA Q heads → MKL GEMM KQ → online softmax →
+//   MKL GEMM VKQ → accumulate → normalize → scatter to dst
+// ---------------------------------------------------------------------------
+void ggml_sycl_flash_attn_ext_mkl(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
+
+    const ggml_tensor * Q    = dst->src[0];
+    const ggml_tensor * K    = dst->src[1];
+    const ggml_tensor * V    = dst->src[2];
+    const ggml_tensor * mask = dst->src[3];
+    ggml_tensor * KQV = dst;
+
+    GGML_ASSERT(Q->type == GGML_TYPE_F32);
+    GGML_ASSERT(KQV->type == GGML_TYPE_F32);
+
+    // --- Op params ---
+    float scale = 1.0f, max_bias = 0.0f, logit_softcap = 0.0f;
+    memcpy(&scale,         (const float *)KQV->op_params + 0, sizeof(float));
+    memcpy(&max_bias,      (const float *)KQV->op_params + 1, sizeof(float));
+    memcpy(&logit_softcap, (const float *)KQV->op_params + 2, sizeof(float));
+
+    const float q_scale = scale;
+
+    // --- Dimensions ---
+    const int DKQ        = (int)K->ne[0];
+    const int DV         = (int)V->ne[0];
+    const int n_queries  = (int)Q->ne[1];
+    const int n_q_heads  = (int)Q->ne[2];
+    const int n_kv_heads = (int)K->ne[2];
+    const int n_batch    = (int)Q->ne[3];
+    const int n_kv       = (int)K->ne[1];
+    const int gqa_ratio  = n_q_heads / n_kv_heads;
+    const int n_query_rows = n_queries * gqa_ratio;
+
+    GGML_ASSERT(n_q_heads % n_kv_heads == 0);
+    GGML_ASSERT(max_bias == 0.0f);  // ALiBi not supported
+    GGML_ASSERT(Q->ne[3] == K->ne[3] || K->ne[3] == 1);
+
+    const int chunk_size = std::min(MKL_FA_CHUNK_SIZE_KV, n_kv);
+
+    // Query rows are processed in tiles of q_tile_rows so the score buffers
+    // (KQ_f32/S_f16 = q_tile_rows * chunk_size) stay bounded regardless of
+    // batch size. n_query_rows <= Q_TILE is a single tile (no extra work).
+    static int q_tile_env = ggml_sycl_get_env("GGML_SYCL_MKL_FA_Q_TILE", MKL_FA_Q_TILE);
+    const int q_tile_rows = std::max(1, std::min(q_tile_env, n_query_rows));
+
+    const int64_t wg_size = MKL_FA_WG_SIZE;
+
+    // --- Debug output (gated by GGML_SYCL_MKL_FA_DEBUG=1) ---
+    static int mkl_call_count = 0;
+    mkl_call_count++;
+    static int mkl_debug = ggml_sycl_get_env("GGML_SYCL_MKL_FA_DEBUG", 0);
+    const bool do_print = (mkl_debug == 1);
+
+    const int64_t q_row_stride  = Q->nb[1] / sizeof(float);
+    const int64_t q_head_stride = Q->nb[2] / sizeof(float);
+
+    const bool V_is_K_view = V->view_src
+        && (V->view_src == K || (V->view_src == K->view_src
+            && V->view_offs == K->view_offs));
+
+    // Early interleaved detection for debug output.
+    // True interleaved detection happens after dequant (nb12_fp16 == nb11_fp16),
+    // but we can pre-detect on the original tensor strides.
+    const bool k_early_interleaved =
+        ((int64_t)K->ne[1] * K->nb[1] != K->nb[2]);
+    const bool v_early_interleaved =
+        !V_is_K_view && ((int64_t)V->ne[1] * V->nb[1] != V->nb[2]);
+
+    if (do_print) {
+        GGML_LOG_INFO("[MKL-FA] #%d D=%d DV=%d n_q=%d n_kv=%d "
+                "n_qh=%d n_kvh=%d gqa=%d batch=%d K=%s V=%s "
+                "chunk=%d buf=%.1fMB%s%s\n",
+                mkl_call_count, DKQ, DV, n_queries, n_kv,
+                n_q_heads, n_kv_heads, gqa_ratio, n_batch,
+                ggml_type_name(K->type), ggml_type_name(V->type),
+                chunk_size,
+                (double)((int64_t)n_query_rows * chunk_size * sizeof(float))
+                    / (1024.0 * 1024.0),
+                k_early_interleaved ? " K_ILV" : "",
+                v_early_interleaved ? " V_ILV" : "");
+        GGML_LOG_INFO("[MKL-FA] #%d Q-nb1=%lld Q-nb2=%lld "
+                "q_rs=%lld q_hs=%lld dst_rs=%lld dst_hs=%lld\n",
+                mkl_call_count,
+                (long long)Q->nb[1], (long long)Q->nb[2],
+                (long long)q_row_stride, (long long)q_head_stride,
+                (long long)(KQV->nb[1] / sizeof(float)),
+                (long long)(KQV->nb[2] / sizeof(float)));
+    }
+
+    // --- Stream and allocators ---
+    dpct::queue_ptr stream = ctx.stream();
+
+#define MKL_TAKE_TIME(t0)  auto t0 = std::chrono::steady_clock::now()
+#define MKL_ACCUM(acc, t0) do { if (do_print) { \
+    acc += (int64_t)std::chrono::duration_cast \
+    <std::chrono::microseconds>(std::chrono::steady_clock::now() - (t0)).count(); \
+} } while(0)
+
+    int64_t gemm_kq_time_us  = 0;
+    int64_t gemm_vkq_time_us = 0;
+    int64_t softmax_time_us  = 0;
+    int64_t dequant_time_us  = 0;
+
+    MKL_TAKE_TIME(t_deq);
+
+    // --- K/V dequant descriptors ---
+    // Dequant is done per-chunk inside the KV loop (footprint independent of
+    // context). Output is always dense row-major fp16 [this_chunk x D], lda=D.
+    // Interleaved detection: ne[1]*nb[1] != nb[2] means heads are interleaved.
+    const bool k_interleaved =
+        ((int64_t)K->ne[1] * K->nb[1] != K->nb[2]) && K->ne[2] > 1;
+    const bool v_interleaved =
+        ((int64_t)V->ne[1] * V->nb[1] != V->nb[2]) && V->ne[2] > 1;
+
+    const mkl_fa_kv_desc K_desc = mkl_fa_make_desc(K, k_interleaved, n_kv_heads);
+    const mkl_fa_kv_desc V_desc = V_is_K_view
+        ? K_desc : mkl_fa_make_desc(V, v_interleaved, n_kv_heads);
+
+    MKL_ACCUM(dequant_time_us, t_deq);
+
+    // --- Resolve mask pointers ---
+    const sycl::half * mask_data = nullptr;
+    int64_t mask_head_stride = 0;
+    int64_t mask_row_stride  = 0;
+    int     mask_n_heads     = 0;
+
+    if (mask) {
+        // Use actual fp16 device size (2 bytes), NOT sizeof(sycl::half)
+        // which may be 4 on the host in oneAPI.
+        mask_head_stride = mask->nb[2] / 2;
+        mask_row_stride  = mask->nb[1] / 2;
+        mask_n_heads     = (int)mask->ne[2];
+    }
+
+    // --- Allocate intermediates from pool ---
+    ggml_sycl_pool & pool = ctx.pool();
+
+    ggml_sycl_pool_alloc<float>      KQ_f32(pool);      // [q_tile_rows x chunk]
+    ggml_sycl_pool_alloc<sycl::half> S_f16(pool);       // [q_tile_rows x chunk]
+    ggml_sycl_pool_alloc<float>      VKQ_chunk(pool);   // [q_tile_rows x DV]
+    ggml_sycl_pool_alloc<float>      VKQ_accum(pool);   // [n_query_rows x DV] (full)
+    ggml_sycl_pool_alloc<float>      KQ_max(pool);      // [n_query_rows] (full)
+    ggml_sycl_pool_alloc<float>      KQ_sum(pool);      // [n_query_rows] (full)
+    ggml_sycl_pool_alloc<sycl::half> Q_head_f16(pool);  // [n_query_rows x DKQ] (full)
+    ggml_sycl_pool_alloc<sycl::half> K_chunk_f16(pool); // [chunk x DKQ] (per-chunk dequant)
+    ggml_sycl_pool_alloc<sycl::half> V_chunk_f16(pool); // [chunk x DV] (per-chunk dequant)
+
+    KQ_f32.alloc((size_t)q_tile_rows * chunk_size);
+    S_f16.alloc((size_t)q_tile_rows * chunk_size);
+    VKQ_chunk.alloc((size_t)q_tile_rows * DV);
+    VKQ_accum.alloc((size_t)n_query_rows * DV);
+    KQ_max.alloc(n_query_rows);
+    KQ_sum.alloc(n_query_rows);
+    Q_head_f16.alloc((size_t)n_query_rows * DKQ);
+    K_chunk_f16.alloc((size_t)chunk_size * DKQ);
+
+    sycl::half * V_chunk_f16_ptr;
+    if (V_is_K_view) {
+        V_chunk_f16_ptr = K_chunk_f16.ptr;   // V aliases K (DV == DKQ)
+    } else {
+        V_chunk_f16.alloc((size_t)chunk_size * DV);
+        V_chunk_f16_ptr = V_chunk_f16.ptr;
+    }
+
+    sycl::half * Q_head_f16_ptr  = Q_head_f16.ptr;
+    float      * KQ_f32_ptr      = KQ_f32.ptr;
+    sycl::half * S_f16_ptr       = S_f16.ptr;
+    float      * VKQ_chunk_ptr   = VKQ_chunk.ptr;
+    float      * VKQ_accum_ptr   = VKQ_accum.ptr;
+    float      * KQ_max_ptr      = KQ_max.ptr;
+    float      * KQ_sum_ptr      = KQ_sum.ptr;
+    sycl::half * K_chunk_f16_ptr = K_chunk_f16.ptr;
+
+    const float alpha = 1.0f;
+    const float beta  = 0.0f;
+
+    for (int ib = 0; ib < n_batch; ib++) {
+        const float * Q_batch = (const float *)Q->data
+            + ib * (Q->nb[3] / sizeof(float));
+        float * dst_batch = (float *)KQV->data
+            + ib * (KQV->nb[3] / sizeof(float));
+
+        const sycl::half * mask_batch = nullptr;
+        if (mask) {
+            int m_batch = (mask->ne[3] > 1) ? ib : 0;
+            mask_batch = (const sycl::half *)mask->data
+                + m_batch * (mask->nb[3] / 2);  // 2 = actual fp16 device size
+        }
+
+        for (int ikvh = 0; ikvh < n_kv_heads; ikvh++) {
+            int kvh_base_head = ikvh * gqa_ratio;
+
+            // 1. Pack all GQA Q heads into fp16 (full n_query_rows)
+            mkl_fa_pack_q_fp16(stream,
+                Q_head_f16_ptr, Q_batch,
+                n_queries, n_query_rows, DKQ,
+                gqa_ratio, kvh_base_head,
+                q_scale, q_row_stride, q_head_stride, wg_size);
+
+            // 2. Initialize softmax state (full n_query_rows)
+            mkl_fa_init_softmax_state(stream,
+                KQ_max_ptr, KQ_sum_ptr, VKQ_accum_ptr,
+                n_query_rows, DV, wg_size);
+
+            // Sync before MKL GEMM (MKL may use an internal queue)
+            stream->wait();
+
+            // 3. KV chunk loop (OUTER): dequant each chunk once, then tile queries.
+            for (int chunk_start = 0; chunk_start < n_kv; chunk_start += chunk_size) {
+                int this_chunk = std::min(chunk_size, n_kv - chunk_start);
+
+                // 3a. Dequant this KV chunk to dense fp16 (once per chunk)
+                {
+                    MKL_TAKE_TIME(t0);
+                    mkl_fa_dequant_chunk(stream, K_desc, KQV,
+                        K_chunk_f16_ptr, ikvh, chunk_start, this_chunk);
+                    if (!V_is_K_view) {
+                        mkl_fa_dequant_chunk(stream, V_desc, KQV,
+                            V_chunk_f16_ptr, ikvh, chunk_start, this_chunk);
+                    }
+                    stream->wait();  // dequant must be ready before MKL GEMM
+                    MKL_ACCUM(dequant_time_us, t0);
+                }
+
+                // 3b. Query tile loop (INNER) — bounds KQ_f32/S_f16 footprint.
+                for (int q0 = 0; q0 < n_query_rows; q0 += q_tile_rows) {
+                    int q_rows = std::min(q_tile_rows, n_query_rows - q0);
+
+                    // GEMM: KQ = Q_tile × K_chunk^T
+                    {
+                        MKL_TAKE_TIME(t0);
+                        sycl::event ev = gemm(*stream,
+                            transpose::trans, transpose::nontrans,
+                            this_chunk, q_rows, DKQ,
+                            alpha,
+                            K_chunk_f16_ptr, DKQ,
+                            Q_head_f16_ptr + (int64_t)q0 * DKQ, DKQ,
+                            beta,
+                            KQ_f32_ptr, this_chunk);
+                        try { ev.wait_and_throw(); } catch (sycl::exception & e) {
+                            GGML_LOG_INFO("[MKL-FA] GEMM KQ: %s\n", e.what());
+                            GGML_ABORT("MKL GEMM KQ failed");
+                        }
+                        MKL_ACCUM(gemm_kq_time_us, t0);
+                    }
+                    // Online softmax over this chunk for this query tile
+                    {
+                        MKL_TAKE_TIME(t0);
+                        mkl_fa_online_softmax_chunk(stream,
+                            KQ_f32_ptr, S_f16_ptr,
+                            KQ_max_ptr, KQ_sum_ptr, VKQ_accum_ptr,
+                            q0, q_rows, n_queries, DV,
+                            this_chunk, chunk_start,
+                            kvh_base_head, gqa_ratio,
+                            mask_batch, mask_head_stride,
+                            mask_row_stride, mask_n_heads,
+                            logit_softcap, wg_size);
+                        stream->wait();  // S_f16 must be ready for GEMM
+                        MKL_ACCUM(softmax_time_us, t0);
+                    }
+
+                    // GEMM: VKQ_chunk = S × V_chunk
+                    {
+                        MKL_TAKE_TIME(t0);
+                        sycl::event ev = gemm(*stream,
+                            transpose::nontrans, transpose::nontrans,
+                            DV, q_rows, this_chunk,
+                            alpha,
+                            V_chunk_f16_ptr, DV,
+                            S_f16_ptr, this_chunk,
+                            beta,
+                            VKQ_chunk_ptr, DV);
+                        try { ev.wait_and_throw(); } catch (sycl::exception & e) {
+                            GGML_LOG_INFO("[MKL-FA] GEMM VKQ: %s\n", e.what());
+                            GGML_ABORT("MKL GEMM VKQ failed");
+                        }
+                        MKL_ACCUM(gemm_vkq_time_us, t0);
+                    }
+                    // VKQ_accum[q0..] += VKQ_chunk
+                    {
+                        const int64_t n_total = (int64_t)q_rows * DV;
+                        const int64_t wg = ((n_total + wg_size - 1) / wg_size)
+                            * wg_size;
+                        float * accum = VKQ_accum_ptr + (int64_t)q0 * DV;
+                        stream->submit([&](sycl::handler & cgh) {
+                            cgh.parallel_for(sycl::nd_range<1>(wg, wg_size),
+                                [=](sycl::nd_item<1> item) {
+                                    int64_t i = item.get_global_id(0);
+                                    if (i < n_total) {
+                                        accum[i] += VKQ_chunk_ptr[i];
+                                    }
+                                });
+                        });
+                    }
+                }
+            }
+
+            // 4. Normalize and scatter each GQA head to dst
+            for (int iqg = 0; iqg < gqa_ratio; iqg++) {
+                int     iqh        = kvh_base_head + iqg;
+                int64_t src_offset = (int64_t)iqg * n_queries * DV;
+                mkl_fa_normalize_head(stream,
+                    dst_batch, VKQ_accum_ptr, KQ_sum_ptr,
+                    iqh, n_queries, DV, n_q_heads,
+                    src_offset, wg_size);
+            }
+        }
+    }
+
+#undef MKL_TAKE_TIME
+#undef MKL_ACCUM
+
+    if (do_print) {
+        const int64_t v_chunk_elems = V_is_K_view ? 0 : (int64_t)chunk_size * DV;
+        double total_mb = (double)(
+            (int64_t)q_tile_rows * chunk_size * sizeof(float)      // KQ_f32
+          + (int64_t)q_tile_rows * chunk_size * sizeof(sycl::half) // S_f16
+          + (int64_t)q_tile_rows * DV * sizeof(float)              // VKQ_chunk
+          + (int64_t)n_query_rows * DV * sizeof(float)             // VKQ_accum
+          + (int64_t)n_query_rows * sizeof(float)                  // KQ_max
+          + (int64_t)n_query_rows * sizeof(float)                  // KQ_sum
+          + (int64_t)n_query_rows * DKQ * sizeof(sycl::half)       // Q_head_f16
+          + (int64_t)chunk_size * DKQ * sizeof(sycl::half)         // K_chunk_f16
+          + v_chunk_elems * (int64_t)sizeof(sycl::half)            // V_chunk_f16
+        ) / (1024.0 * 1024.0);
+        GGML_LOG_INFO("[MKL-FA] #%d n_kv=%d n_q=%d q_tile=%d time_us: "
+                "dequant=%lld GEMM_KQ=%lld softmax=%lld GEMM_VKQ=%lld "
+                "buf_mb=%.1f\n",
+                mkl_call_count, n_kv, n_queries, q_tile_rows,
+                (long long)dequant_time_us,
+                (long long)gemm_kq_time_us,
+                (long long)softmax_time_us,
+                (long long)gemm_vkq_time_us,
+                total_mb);
+    }
+}
diff --git a/ggml/src/ggml-sycl/fattn.cpp b/ggml/src/ggml-sycl/fattn.cpp
index 1772b9c85..f8efa53ac 100644
--- a/ggml/src/ggml-sycl/fattn.cpp
+++ b/ggml/src/ggml-sycl/fattn.cpp
@@ -99,8 +99,10 @@ enum best_fattn_kernel {
     BEST_FATTN_KERNEL_VEC      = 100,
     BEST_FATTN_KERNEL_ONEDNN   = 150, // added enum for onednn==150
     BEST_FATTN_KERNEL_TILE     = 200,
+    BEST_FATTN_KERNEL_MKL      = 300,
 };
 
+
 static best_fattn_kernel ggml_sycl_get_best_fattn_kernel(const int device, const ggml_tensor * dst) {
     GGML_UNUSED(device);
 #ifndef SYCL_FLASH_ATTN
@@ -115,6 +117,7 @@ static best_fattn_kernel ggml_sycl_get_best_fattn_kernel(const int device, const
     const ggml_tensor * K     = dst->src[1];
     const ggml_tensor * V     = dst->src[2];
     const ggml_tensor * mask  = dst->src[3];
+    const ggml_tensor * sinks = dst->src[4];
 
     const int gqa_ratio = Q->ne[2] / K->ne[2];
     GGML_ASSERT(Q->ne[2] % K->ne[2] == 0);
@@ -122,7 +125,49 @@ static best_fattn_kernel ggml_sycl_get_best_fattn_kernel(const int device, const
     float max_bias = 0.0f;
     memcpy(&max_bias, (const float *) KQV->op_params + 1, sizeof(float));
 
+    float logit_softcap = 0.0f;
+    memcpy(&logit_softcap, (const float *) KQV->op_params + 2, sizeof(float));
+
     bool gqa_opt_applies = gqa_ratio >= 2 && mask && max_bias == 0.0f && K->ne[1] % FATTN_KQ_STRIDE == 0;
+
+    // MKL path: XMX-accelerated GEMM for prompt processing (all KV cache types).
+    // The MKL kernel converts non-F16 K/V to F16 via to_fp16_sycl before GEMM,
+    // so quantized, F16, BF16, and F32 caches all benefit from XMX acceleration.
+    // Activates automatically when flash-attn is enabled (--flash-attn on or -fa)
+    // and n_kv >= 1024. Falls through to TILE/VEC for ALiBi, logit softcap,
+    // and mismatched batch dimensions (unsupported by the MKL kernel).
+    // Set GGML_SYCL_ENABLE_MKL_FA=0 to force TILE/VEC path for A/B testing.
+    // Example: GGML_SYCL_ENABLE_MKL_FA=0 llama-cli -m model.gguf -fa -ngl 99 ...
+    // Note: MKL GEMM calls are incompatible with SYCL graph capture replay.
+    static int mkl_enable = ggml_sycl_get_env("GGML_SYCL_ENABLE_MKL_FA", 1);
+    // MKL is validated for the mainstream GQA envelope: grouped-query
+    // (gqa_ratio >= 2), head_dim a multiple of 64 in [64,512] with matching
+    // K/V head size, mask, no sinks/ALiBi/softcap. Gemma's global layers use
+    // head_dim 512, so the cap must include it. Head sizes not a multiple of
+    // 64 (72/80/96), MHA (gqa_ratio == 1), and MLA (DKQ != DV, e.g. 576/512)
+    // fall through to TILE/VEC; see follow-up work.
+    if (mkl_enable == 1 && mask && !sinks && gqa_ratio >= 2 &&
+        Q->ne[0] >= 64 && Q->ne[0] <= 512 && Q->ne[0] % 64 == 0 &&
+        Q->ne[0] == V->ne[0] &&
+        Q->ne[1] >= 32 && K->ne[1] >= 1024 &&
+        max_bias == 0.0f && logit_softcap == 0.0f &&
+        (Q->ne[3] == K->ne[3] || K->ne[3] == 1)) {
+        // F16 K/V strides must be a multiple of ne[0]*2 (the natural row size
+        // in bytes). This passes both dense (nb1 == ne0*2) and interleaved
+        // (nb1 == H * ne0*2). Only pathological test strides like nb1=32 or
+        // nb1=75 for ne0=40 fall through to TILE.
+        bool kv_strides_ok = true;
+        for (const ggml_tensor * t : {K, V}) {
+            if (t->type == GGML_TYPE_F16 && t->nb[1] % (t->ne[0] * 2) != 0) {
+                kv_strides_ok = false;
+                break;
+            }
+        }
+        if (kv_strides_ok) {
+            return BEST_FATTN_KERNEL_MKL;
+        }
+    }
+
     for (const ggml_tensor * t : {Q, K, V, mask}) {
         if (t == nullptr || ggml_is_quantized(t->type)) {
             continue;
@@ -216,6 +261,37 @@ static best_fattn_kernel ggml_sycl_get_best_fattn_kernel(const int device, const
 
 void ggml_sycl_flash_attn_ext(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
     ggml_sycl_set_device(ctx.device);
+
+    // n_kv watchdog: log when n_kv differs from the last FA call with
+    // the same D — helps detect cache-truncation issues.
+    static int nkv_debug = ggml_sycl_get_env("GGML_SYCL_MKL_FA_DEBUG", 0);
+    if (nkv_debug == 1) {
+        const ggml_tensor * K_dbg = dst->src[1];
+        const ggml_tensor * V_dbg = dst->src[2];
+        static int64_t last_nkv_d256 = 0, last_nkv_d512 = 0;
+        static int fa_call_seq = 0;
+        fa_call_seq++;
+        int64_t cur_nkv = K_dbg->ne[1];
+        int Dk = (int)K_dbg->ne[0];
+        const char * kname = "TILE";
+        best_fattn_kernel k = ggml_sycl_get_best_fattn_kernel(ctx.device, dst);
+        if (k == BEST_FATTN_KERNEL_MKL)  kname = "MKL";
+        if (k == BEST_FATTN_KERNEL_VEC)  kname = "VEC";
+        int64_t delta = 0;
+        if (Dk == 256) {
+            delta = cur_nkv - last_nkv_d256;
+            last_nkv_d256 = cur_nkv;
+        } else if (Dk == 512) {
+            delta = cur_nkv - last_nkv_d512;
+            last_nkv_d512 = cur_nkv;
+        }
+        GGML_LOG_INFO("[FA-DISP] #%d %s D=%d n_kv=%lld delta=%lld "
+                "V_ne1=%lld\n",
+                fa_call_seq, kname, Dk,
+                (long long)cur_nkv, (long long)delta,
+                (long long)V_dbg->ne[1]);
+    }
+
     switch (ggml_sycl_get_best_fattn_kernel(ggml_sycl_get_device(), dst)) {
         case BEST_FATTN_KERNEL_NONE:
             GGML_ABORT("Not support Flash-Attention");
@@ -232,6 +308,51 @@ void ggml_sycl_flash_attn_ext(ggml_backend_sycl_context & ctx, ggml_tensor * dst
         case BEST_FATTN_KERNEL_VEC:
             ggml_sycl_flash_attn_ext_vec(ctx, dst);
             break;
+        case BEST_FATTN_KERNEL_MKL:
+            ggml_sycl_flash_attn_ext_mkl(ctx, dst);
+            break;
+    }
+
+    // --- Output fingerprint (GGML_SYCL_MKL_FA_DIAG=1) ---
+    // Copy first 64 float output values to host for fingerprinting.
+    // Compare MKL vs TILE (GGML_SYCL_ENABLE_MKL_FA=0) to detect divergence.
+    // Only fingerprints the first 6 FA calls with n_kv >= 1024.
+    static int fa_diag = ggml_sycl_get_env("GGML_SYCL_MKL_FA_DIAG", 0);
+    static int fa_diag_count = 0;
+    if (fa_diag == 1 && fa_diag_count < 6) {
+        const ggml_tensor * K_diag = dst->src[1];
+        const ggml_tensor * V_diag = dst->src[2];
+        const ggml_tensor * Q_diag = dst->src[0];
+        if (K_diag->ne[1] >= 1024) {
+            fa_diag_count++;
+            float diag_buf[64];
+            dpct::queue_ptr q = ctx.stream();
+            q->memcpy(diag_buf, dst->data, 64 * sizeof(float));
+            q->wait();
+            const char * kname = "???";
+            best_fattn_kernel kb = ggml_sycl_get_best_fattn_kernel(ctx.device, dst);
+            if (kb == BEST_FATTN_KERNEL_MKL) kname = "MKL";
+            if (kb == BEST_FATTN_KERNEL_TILE) kname = "TILE";
+            if (kb == BEST_FATTN_KERNEL_VEC) kname = "VEC";
+            GGML_LOG_INFO("[FA-DIAG] #%d %s D=%d n_kv=%lld n_q=%lld "
+                    "n_qh=%lld n_kvh=%lld K=%s V=%s "
+                    "nb1=%zu nb2=%zu first 64 floats:\n",
+                    fa_diag_count, kname,
+                    (int)K_diag->ne[0], (long long)K_diag->ne[1],
+                    (long long)Q_diag->ne[1],
+                    (long long)Q_diag->ne[2], (long long)K_diag->ne[2],
+                    ggml_type_name(K_diag->type),
+                    ggml_type_name(V_diag->type),
+                    K_diag->nb[1], K_diag->nb[2]);
+            for (int i = 0; i < 64; i += 8) {
+                GGML_LOG_INFO("  [%2d] %08x %08x %08x %08x %08x %08x %08x %08x\n",
+                        i,
+                        *(unsigned *)&diag_buf[i+0], *(unsigned *)&diag_buf[i+1],
+                        *(unsigned *)&diag_buf[i+2], *(unsigned *)&diag_buf[i+3],
+                        *(unsigned *)&diag_buf[i+4], *(unsigned *)&diag_buf[i+5],
+                        *(unsigned *)&diag_buf[i+6], *(unsigned *)&diag_buf[i+7]);
+            }
+        }
     }
 }
 
diff --git a/ggml/src/ggml-sycl/fattn.hpp b/ggml/src/ggml-sycl/fattn.hpp
index f2a8ffc97..c093970a3 100644
--- a/ggml/src/ggml-sycl/fattn.hpp
+++ b/ggml/src/ggml-sycl/fattn.hpp
@@ -19,4 +19,6 @@ void ggml_sycl_flash_attn_ext(ggml_backend_sycl_context & ctx, ggml_tensor * dst
 
 bool ggml_sycl_flash_attn_ext_supported(int device, const ggml_tensor * dst);
 
+void ggml_sycl_flash_attn_ext_mkl(ggml_backend_sycl_context & ctx, ggml_tensor * dst);
+
 #endif // GGML_SYCL_FATTN_HPP
diff --git a/tests/test-backend-ops.cpp b/tests/test-backend-ops.cpp
index 162bfc98c..e101ca390 100644
--- a/tests/test-backend-ops.cpp
+++ b/tests/test-backend-ops.cpp
@@ -9513,6 +9513,18 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
         }
     }
 
+    // prefill-shaped cases with long KV (nb >= 32, kv >= 1024): covers the
+    // XMX/GEMM-accelerated SYCL FA path which only activates for these shapes.
+    for (int kv : { 1024, 2048, }) {
+        for (int hs : { 64, 128, 256, }) {
+            for (int nb : { 32, 64, }) {
+                for (ggml_type type_KV : { GGML_TYPE_F16, GGML_TYPE_Q8_0, GGML_TYPE_Q4_0, }) {
+                    test_cases.emplace_back(new test_flash_attn_ext(hs, hs, 8, {4, 1}, kv, nb, true, false, 0, 0, GGML_PREC_F32, type_KV, type_KV));
+                }
+            }
+        }
+    }
+
     for (int hsk : { 40, 64, 72, 80, 96, 128, 192, 256, 320, 512, 576 }) {
         for (int hsv : { 40, 64, 72, 80, 96, 128, 192, 256, 512 }) {
             if (hsk != 192 && hsk != 320 && hsk != 576 && hsk != hsv) continue;

From 15e755f30d3deb75fea2c9cc8cd129a8d7cf634d Mon Sep 17 00:00:00 2001
From: David Friehs <david@friehs.info>
Date: Fri, 31 Jul 2026 10:15:44 +0200
Subject: [PATCH 29/30] cuda: extract Q2_0 elements via __byte_perm (#25603)

---
 ggml/src/ggml-cuda/mmq-load-tiles.cuh | 48 +++++++++----------------
 ggml/src/ggml-cuda/vecdotq.cuh        | 52 +++++++++------------------
 2 files changed, 32 insertions(+), 68 deletions(-)

diff --git a/ggml/src/ggml-cuda/mmq-load-tiles.cuh b/ggml/src/ggml-cuda/mmq-load-tiles.cuh
index 8c9070c6b..8ed704c28 100644
--- a/ggml/src/ggml-cuda/mmq-load-tiles.cuh
+++ b/ggml/src/ggml-cuda/mmq-load-tiles.cuh
@@ -130,43 +130,27 @@ template <ggml_type type, int J, bool fallback> static __device__ __forceinline_
         }
 
         const block_q2_0 * bxi = (const block_q2_0 *) x + kbx0 + i*stride + kbx;
-        // Each 32-element chunk occupies 8 bytes of qs (32 elements * 2 bits = 64 bits)
-        const int qs_offset = 8*kqsx;
-        const int qs0 = bxi->qs[qs_offset + 0] | (bxi->qs[qs_offset + 1] << 8) |
-                        (bxi->qs[qs_offset + 2] << 16) | (bxi->qs[qs_offset + 3] << 24);
-        const int qs1 = bxi->qs[qs_offset + 4] | (bxi->qs[qs_offset + 5] << 8) |
-                        (bxi->qs[qs_offset + 6] << 16) | (bxi->qs[qs_offset + 7] << 24);
-
-        // Unpack 32 2-bit codes into 8 int32s, each holding 4 signed int8s in {-1,0,1,2}.
-        int unpacked_bytes[8];
-#pragma unroll
-        for (int j = 0; j < 4; ++j) {
-            const int shift = j * 8;
-            const int codes = (qs0 >> shift) & 0xFF;
-            const int c0 = ((codes >> 0) & 0x3) - 1;
-            const int c1 = ((codes >> 2) & 0x3) - 1;
-            const int c2 = ((codes >> 4) & 0x3) - 1;
-            const int c3 = ((codes >> 6) & 0x3) - 1;
-            unpacked_bytes[j] = (c0 & 0xFF) | ((c1 & 0xFF) << 8) | ((c2 & 0xFF) << 16) | ((c3 & 0xFF) << 24);
-        }
-#pragma unroll
-        for (int j = 0; j < 4; ++j) {
-            const int shift = j * 8;
-            const int codes = (qs1 >> shift) & 0xFF;
-            const int c0 = ((codes >> 0) & 0x3) - 1;
-            const int c1 = ((codes >> 2) & 0x3) - 1;
-            const int c2 = ((codes >> 4) & 0x3) - 1;
-            const int c3 = ((codes >> 6) & 0x3) - 1;
-            unpacked_bytes[4 + j] = (c0 & 0xFF) | ((c1 & 0xFF) << 8) | ((c2 & 0xFF) << 16) | ((c3 & 0xFF) << 24);
-        }
+        const int16_t    * qxi = (const int16_t *) bxi->qs + kqsx * 4;
 
         const int dst_offset = kbx*(scale_entries_per_block*QI8_0) + kqsx*QI8_0;
+
 #pragma unroll
-        for (int j = 0; j < 8; ++j) {
+        for (int j = 0; j < 4; ++j) {
+            const int q  = qxi[j];
+
+            // unpack even and odd crumbs into byte values
+            const int qe = __byte_perm(0x020100FF, 0x020100FF, q >> 0);
+            const int qo = __byte_perm(0x020100FF, 0x020100FF, q >> 2);
+            // unshuffle values
+            const int qx = __byte_perm(qe, qo, 0x5140);
+            const int qy = __byte_perm(qe, qo, 0x7362);
+
 #if defined(AMD_MFMA_AVAILABLE) || defined(TURING_MMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
-            x_qs[i*sram_stride           + dst_offset + j] = unpacked_bytes[j];
+            x_qs[i*sram_stride           + dst_offset + j*2+0] = qx;
+            x_qs[i*sram_stride           + dst_offset + j*2+1] = qy;
 #else
-            x_qs[i*(2*MMQ_TILE_NE_K + 1) + dst_offset + j] = unpacked_bytes[j];
+            x_qs[i*(2*MMQ_TILE_NE_K + 1) + dst_offset + j*2+0] = qx;
+            x_qs[i*(2*MMQ_TILE_NE_K + 1) + dst_offset + j*2+1] = qy;
 #endif // defined(AMD_MFMA_AVAILABLE) || defined(TURING_MMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
         }
     }
diff --git a/ggml/src/ggml-cuda/vecdotq.cuh b/ggml/src/ggml-cuda/vecdotq.cuh
index 36d89c89e..0f039c735 100644
--- a/ggml/src/ggml-cuda/vecdotq.cuh
+++ b/ggml/src/ggml-cuda/vecdotq.cuh
@@ -734,48 +734,28 @@ static __device__ __forceinline__ float vec_dot_q2_0_q8_1(
     // Q8_1: 32 elements per block with individual scales
     // iqs selects which of the 2 chunks of 32 elements to process (0-1)
 
-    const float d2 = bq2_0->d;
+    const float     d2 = bq2_0->d;
+    const int16_t * qs = (const int16_t *) bq2_0->qs + iqs * 4;
 
     // Process only the chunk specified by iqs
     const block_q8_1 * bq8_1_chunk = bq8_1 + iqs;
 
-    // Load 64 bits (8 bytes) for this chunk from Q2_0: bytes [8*iqs, 8*iqs+8)
-    const int offset = iqs * 8;
-    const int v0 = bq2_0->qs[offset + 0] | (bq2_0->qs[offset + 1] << 8) |
-                   (bq2_0->qs[offset + 2] << 16) | (bq2_0->qs[offset + 3] << 24);
-    const int v1 = bq2_0->qs[offset + 4] | (bq2_0->qs[offset + 5] << 8) |
-                   (bq2_0->qs[offset + 6] << 16) | (bq2_0->qs[offset + 7] << 24);
-
-    // Unpack 32 2-bit codes into 8 int32s, each holding 4 signed int8 symbols in {-1,0,1,2}.
-    // Stored code c in {0,1,2,3} -> symbol s = c - 1.
-    int vi_bytes[8];
-#pragma unroll
-    for (int j = 0; j < 4; ++j) {
-        const int shift = j * 8;
-        const int codes = (v0 >> shift) & 0xFF;
-        const int c0 = ((codes >> 0) & 0x3) - 1;
-        const int c1 = ((codes >> 2) & 0x3) - 1;
-        const int c2 = ((codes >> 4) & 0x3) - 1;
-        const int c3 = ((codes >> 6) & 0x3) - 1;
-        vi_bytes[j] = (c0 & 0xFF) | ((c1 & 0xFF) << 8) | ((c2 & 0xFF) << 16) | ((c3 & 0xFF) << 24);
-    }
-#pragma unroll
-    for (int j = 0; j < 4; ++j) {
-        const int shift = j * 8;
-        const int codes = (v1 >> shift) & 0xFF;
-        const int c0 = ((codes >> 0) & 0x3) - 1;
-        const int c1 = ((codes >> 2) & 0x3) - 1;
-        const int c2 = ((codes >> 4) & 0x3) - 1;
-        const int c3 = ((codes >> 6) & 0x3) - 1;
-        vi_bytes[4 + j] = (c0 & 0xFF) | ((c1 & 0xFF) << 8) | ((c2 & 0xFF) << 16) | ((c3 & 0xFF) << 24);
-    }
-
-    // Compute dot product for this 32-element chunk
     int sumi = 0;
 #pragma unroll
-    for (int j = 0; j < 8; ++j) {
-        const int u = get_int_b4(bq8_1_chunk->qs, j);
-        sumi = ggml_cuda_dp4a(vi_bytes[j], u, sumi);
+    for (int j = 0; j < 4; ++j) {
+        const int q  = qs[j];
+        const int u  = get_int_b4(bq8_1_chunk->qs, j*2+0);
+        const int v  = get_int_b4(bq8_1_chunk->qs, j*2+1);
+
+        // unpack even and odd crumbs into byte values
+        const int qe = __byte_perm(0x020100FF, 0x020100FF, q >> 0);
+        const int qo = __byte_perm(0x020100FF, 0x020100FF, q >> 2);
+        // unshuffle values
+        const int qx = __byte_perm(qe, qo, 0x5140);
+        const int qy = __byte_perm(qe, qo, 0x7362);
+
+        sumi = ggml_cuda_dp4a(u, qx, sumi);
+        sumi = ggml_cuda_dp4a(v, qy, sumi);
     }
 
     // Apply Q2_0's single scale and this chunk's Q8_1 scale

From 000547513f1530346ecd163db8b3e13962949961 Mon Sep 17 00:00:00 2001
From: Ruixiang Wang <wangruixiang07@outlook.com>
Date: Fri, 31 Jul 2026 10:16:17 +0200
Subject: [PATCH 30/30] server: correct accepted tokens when need draft token
 replay (#26320)

* spec: correct accepted tokens when need draft token replay

* cont : naming

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
---
 tools/server/server-context.cpp | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/tools/server/server-context.cpp b/tools/server/server-context.cpp
index 0f7ada435..4655b518e 100644
--- a/tools/server/server-context.cpp
+++ b/tools/server/server-context.cpp
@@ -212,6 +212,7 @@ struct server_slot {
     llama_tokens spec_prompt;
     std::vector<int32_t> spec_i_batch;
     common_prompt_checkpoint spec_ckpt;
+    bool spec_is_replay = false;
 
     // TODO: move members that belong to the task (such as `generated_text`, `has_new_line`) to task_results_state
     //       see https://github.com/ggml-org/llama.cpp/pull/18283#issuecomment-3710175837
@@ -332,6 +333,8 @@ struct server_slot {
     void reset() {
         SLT_DBG(*this, "%s", "\n");
 
+        spec_is_replay = false;
+
         n_prompt_tokens_cache = 0;
 
         last_nl_pos    = 0;
@@ -3875,6 +3878,7 @@ private:
                         }
 
                         // partial acceptance is not supported by the context -> truncate the draft and restore the state
+                        slot.spec_is_replay = true;
                         slot.spec_draft = std::move(accepted);
 
                         const auto & ckpt = slot.spec_ckpt;
@@ -3909,16 +3913,22 @@ private:
 
             const auto ids = std::move(slot.spec_draft);
 
+            size_t n_accepted = ids.size() - 1;
+            if (slot.spec_is_replay && n_accepted > 0) {
+                n_accepted--;
+            }
+            slot.spec_is_replay = false;
+
             slot.t_token_generation = std::max<int64_t>(1, t_now - slot.t_start_generation) / 1e3;
 
             // update how many tokens out of those tested were accepted
-            slot.n_draft_accepted += ids.size() - 1;
+            slot.n_draft_accepted += n_accepted;
             slot.n_draft_verif_steps += 1;
 
             if (slot.n_accepted_per_pos.empty()) {
                 slot.n_accepted_per_pos.resize(common_speculative_n_max(&params_base.speculative), 0);
             }
-            for (size_t i = 0; i < ids.size() - 1 && i < slot.n_accepted_per_pos.size(); ++i) {
+            for (size_t i = 0; i < n_accepted && i < slot.n_accepted_per_pos.size(); ++i) {
                 slot.n_accepted_per_pos[i]++;
             }
 
@@ -3954,7 +3964,7 @@ private:
 
             slot.print_timings_tg();
 
-            SLT_DBG(slot, "accepted %d/%d draft tokens, new n_tokens = %d\n", (int) ids.size() - 1, (int) n_draft, slot.prompt.n_tokens());
+            SLT_DBG(slot, "accepted %d/%d draft tokens, new n_tokens = %d\n", (int) n_accepted, (int) n_draft, slot.prompt.n_tokens());
         });
     }