mirror of
https://github.com/kvcache-ai/ktransformers.git
synced 2026-04-28 11:49:51 +00:00
57d14d22bc
* refactor: move legacy code to archive/ directory - Moved ktransformers, csrc, third_party, merge_tensors to archive/ - Moved build scripts and configurations to archive/ - Kept kt-kernel, KT-SFT, doc, and README files in root - Preserved complete git history for all moved files * refactor: restructure repository to focus on kt-kernel and KT-SFT modules * fix README * fix README * fix README * fix README * docs: add performance benchmarks to kt-kernel section Add comprehensive performance data for kt-kernel to match KT-SFT's presentation: - AMX kernel optimization: 21.3 TFLOPS (3.9× faster than PyTorch) - Prefill phase: up to 20× speedup vs baseline - Decode phase: up to 4× speedup - NUMA optimization: up to 63% throughput improvement - Multi-GPU (8×L20): 227.85 tokens/s total throughput with DeepSeek-R1 FP8 Source: https://lmsys.org/blog/2025-10-22-KTransformers/ This provides users with concrete performance metrics for both core modules, making it easier to understand the capabilities of each component. * refactor: improve kt-kernel performance data with specific hardware and models Replace generic performance descriptions with concrete benchmarks: - Specify exact hardware: 8×L20 GPU + Xeon Gold 6454S, Single/Dual-socket Xeon + AMX - Include specific models: DeepSeek-R1-0528 (FP8), DeepSeek-V3 (671B) - Show detailed metrics: total throughput, output throughput, concurrency details - Match KT-SFT presentation style for consistency This provides users with actionable performance data they can use to evaluate hardware requirements and expected performance for their use cases. * fix README * docs: clean up performance table and improve formatting * add pic for README * refactor: simplify .gitmodules and backup legacy submodules - Remove 7 legacy submodules from root .gitmodules (archive/third_party/*) - Keep only 2 active submodules for kt-kernel (llama.cpp, pybind11) - Backup complete .gitmodules to archive/.gitmodules - Add documentation in archive/README.md for researchers who need legacy submodules This reduces initial clone size by ~500MB and avoids downloading unused dependencies. * refactor: move doc/ back to root directory Keep documentation in root for easier access and maintenance. * refactor: consolidate all images to doc/assets/ - Move kt-kernel/assets/heterogeneous_computing.png to doc/assets/ - Remove KT-SFT/assets/ (images already in doc/assets/) - Update KT-SFT/README.md image references to ../doc/assets/ - Eliminates ~7.9MB image duplication - Centralizes all documentation assets in one location * fix pic path for README
240 lines
No EOL
8.2 KiB
YAML
240 lines
No EOL
8.2 KiB
YAML
- match:
|
|
name: "^model.embed_tokens"
|
|
replace:
|
|
class: "default"
|
|
kwargs:
|
|
generate_device: "cpu"
|
|
prefill_device: "cpu"
|
|
|
|
- match:
|
|
name: "^model\\.layers\\.([0-9]|[1][0-4])\\."
|
|
class: ktransformers.models.modeling_deepseek.DeepseekV2YarnRotaryEmbedding
|
|
replace:
|
|
class: ktransformers.operators.RoPE.YarnRotaryEmbedding
|
|
kwargs:
|
|
generate_device: "cuda:0"
|
|
prefill_device: "cuda:0"
|
|
- match:
|
|
name: "^model\\.layers\\.([2][0-9]|[1][5-9])\\."
|
|
class: ktransformers.models.modeling_deepseek.DeepseekV2YarnRotaryEmbedding
|
|
replace:
|
|
class: ktransformers.operators.RoPE.YarnRotaryEmbedding
|
|
kwargs:
|
|
generate_device: "cuda:1"
|
|
prefill_device: "cuda:1"
|
|
- match:
|
|
name: "^model\\.layers\\.([3][0-9]|[4][0-4])\\."
|
|
class: ktransformers.models.modeling_deepseek.DeepseekV2YarnRotaryEmbedding
|
|
replace:
|
|
class: ktransformers.operators.RoPE.YarnRotaryEmbedding
|
|
kwargs:
|
|
generate_device: "cuda:2"
|
|
prefill_device: "cuda:2"
|
|
- match:
|
|
name: "^model\\.layers\\.([5][0-9]|[4][5-9])\\."
|
|
class: ktransformers.models.modeling_deepseek.DeepseekV2YarnRotaryEmbedding
|
|
replace:
|
|
class: ktransformers.operators.RoPE.YarnRotaryEmbedding
|
|
kwargs:
|
|
generate_device: "cuda:3"
|
|
prefill_device: "cuda:3"
|
|
|
|
- match:
|
|
name: "^model\\.layers\\.([0-9]|[1][0-4])\\.(?!self_attn\\.kv_b_proj).*$" # regular expression
|
|
class: torch.nn.Linear # only match modules matching name and class simultaneously
|
|
replace:
|
|
class: ktransformers.operators.linear.KTransformersLinear # optimized Kernel on quantized data types
|
|
kwargs:
|
|
generate_device: "cuda:0"
|
|
prefill_device: "cuda:0"
|
|
generate_op: "KLinearMarlin"
|
|
prefill_op: "KLinearTorch"
|
|
- match:
|
|
name: "^model\\.layers\\.([2][0-9]|[1][5-9])\\.(?!self_attn\\.kv_b_proj).*$" # regular expression
|
|
class: torch.nn.Linear # only match modules matching name and class simultaneously
|
|
replace:
|
|
class: ktransformers.operators.linear.KTransformersLinear # optimized Kernel on quantized data types
|
|
kwargs:
|
|
generate_device: "cuda:1"
|
|
prefill_device: "cuda:1"
|
|
generate_op: "KLinearMarlin"
|
|
prefill_op: "KLinearTorch"
|
|
- match:
|
|
name: "^model\\.layers\\.([3][0-9]|[4][0-4])\\.(?!self_attn\\.kv_b_proj).*$" # regular expression
|
|
class: torch.nn.Linear # only match modules matching name and class simultaneously
|
|
replace:
|
|
class: ktransformers.operators.linear.KTransformersLinear # optimized Kernel on quantized data types
|
|
kwargs:
|
|
generate_device: "cuda:2"
|
|
prefill_device: "cuda:2"
|
|
generate_op: "KLinearMarlin"
|
|
prefill_op: "KLinearTorch"
|
|
- match:
|
|
name: "^model\\.layers\\.([5][0-9]|[4][5-9])\\.(?!self_attn\\.kv_b_proj).*$" # regular expression
|
|
class: torch.nn.Linear # only match modules matching name and class simultaneously
|
|
replace:
|
|
class: ktransformers.operators.linear.KTransformersLinear # optimized Kernel on quantized data types
|
|
kwargs:
|
|
generate_device: "cuda:3"
|
|
prefill_device: "cuda:3"
|
|
generate_op: "KLinearMarlin"
|
|
prefill_op: "KLinearTorch"
|
|
|
|
- match:
|
|
name: "^model\\.layers\\.([0-9]|[1][0-4])\\.mlp$"
|
|
class: ktransformers.models.modeling_deepseek.DeepseekV2MoE
|
|
replace:
|
|
class: ktransformers.operators.experts.KDeepseekV2MoE # mlp module with custom forward function
|
|
kwargs:
|
|
generate_device: "cuda:0"
|
|
prefill_device: "cuda:0"
|
|
- match:
|
|
name: "^model\\.layers\\.([2][0-9]|[1][5-9])\\.mlp$"
|
|
class: ktransformers.models.modeling_deepseek.DeepseekV2MoE
|
|
replace:
|
|
class: ktransformers.operators.experts.KDeepseekV2MoE # mlp module with custom forward function
|
|
kwargs:
|
|
generate_device: "cuda:1"
|
|
prefill_device: "cuda:1"
|
|
- match:
|
|
name: "^model\\.layers\\.([3][0-9]|[4][0-4])\\.mlp$"
|
|
class: ktransformers.models.modeling_deepseek.DeepseekV2MoE
|
|
replace:
|
|
class: ktransformers.operators.experts.KDeepseekV2MoE # mlp module with custom forward function
|
|
kwargs:
|
|
generate_device: "cuda:2"
|
|
prefill_device: "cuda:2"
|
|
- match:
|
|
name: "^model\\.layers\\.([5][0-9]|[4][5-9])\\.mlp$"
|
|
class: ktransformers.models.modeling_deepseek.DeepseekV2MoE
|
|
replace:
|
|
class: ktransformers.operators.experts.KDeepseekV2MoE # mlp module with custom forward function
|
|
kwargs:
|
|
generate_device: "cuda:3"
|
|
prefill_device: "cuda:3"
|
|
|
|
- match:
|
|
name: "^model\\.layers\\.([0-9]|[1][0-4])\\.mlp\\.experts$"
|
|
replace:
|
|
class: ktransformers.operators.experts.KTransformersExperts # custom MoE Kernel with expert paralleism
|
|
kwargs:
|
|
prefill_device: "cuda:0"
|
|
prefill_op: "KExpertsTorch"
|
|
generate_device: "cpu"
|
|
generate_op: "KExpertsCPU"
|
|
out_device: "cuda:0"
|
|
recursive: False # don't recursively inject submodules of this module
|
|
- match:
|
|
name: "^model\\.layers\\.([2][0-9]|[1][5-9])\\.mlp\\.experts$"
|
|
replace:
|
|
class: ktransformers.operators.experts.KTransformersExperts # custom MoE Kernel with expert paralleism
|
|
kwargs:
|
|
prefill_device: "cuda:1"
|
|
prefill_op: "KExpertsTorch"
|
|
generate_device: "cpu"
|
|
generate_op: "KExpertsCPU"
|
|
out_device: "cuda:1"
|
|
recursive: False # don't recursively inject submodules of this module
|
|
- match:
|
|
name: "^model\\.layers\\.([3][0-9]|[4][0-4])\\.mlp\\.experts$"
|
|
replace:
|
|
class: ktransformers.operators.experts.KTransformersExperts # custom MoE Kernel with expert paralleism
|
|
kwargs:
|
|
prefill_device: "cuda:2"
|
|
prefill_op: "KExpertsTorch"
|
|
generate_device: "cpu"
|
|
generate_op: "KExpertsCPU"
|
|
out_device: "cuda:2"
|
|
recursive: False # don't recursively inject submodules of this module
|
|
- match:
|
|
name: "^model\\.layers\\.([5][0-9]|[4][5-9])\\.mlp\\.experts$"
|
|
replace:
|
|
class: ktransformers.operators.experts.KTransformersExperts # custom MoE Kernel with expert paralleism
|
|
kwargs:
|
|
prefill_device: "cuda:3"
|
|
prefill_op: "KExpertsTorch"
|
|
generate_device: "cpu"
|
|
generate_op: "KExpertsCPU"
|
|
out_device: "cuda:3"
|
|
recursive: False # don't recursively inject submodules of this module
|
|
|
|
- match:
|
|
name: "^model\\.layers\\.([0-9]|[1][0-4])\\.self_attn$"
|
|
replace:
|
|
class: ktransformers.operators.attention.KDeepseekV2Attention # optimized MLA implementation
|
|
kwargs:
|
|
generate_device: "cuda:0"
|
|
prefill_device: "cuda:0"
|
|
- match:
|
|
name: "^model\\.layers\\.([2][0-9]|[1][5-9])\\.self_attn$"
|
|
replace:
|
|
class: ktransformers.operators.attention.KDeepseekV2Attention # optimized MLA implementation
|
|
kwargs:
|
|
generate_device: "cuda:1"
|
|
prefill_device: "cuda:1"
|
|
- match:
|
|
name: "^model\\.layers\\.([3][0-9]|[4][0-4])\\.self_attn$"
|
|
replace:
|
|
class: ktransformers.operators.attention.KDeepseekV2Attention # optimized MLA implementation
|
|
kwargs:
|
|
generate_device: "cuda:2"
|
|
prefill_device: "cuda:2"
|
|
- match:
|
|
name: "^model\\.layers\\.([5][0-9]|[4][5-9])\\.self_attn$"
|
|
replace:
|
|
class: ktransformers.operators.attention.KDeepseekV2Attention # optimized MLA implementation
|
|
kwargs:
|
|
generate_device: "cuda:3"
|
|
prefill_device: "cuda:3"
|
|
|
|
- match:
|
|
name: "^model$"
|
|
replace:
|
|
class: "ktransformers.operators.models.KDeepseekV2Model"
|
|
kwargs:
|
|
per_layer_prefill_intput_threshold: 0 # 0 is close layer wise prefill
|
|
transfer_map:
|
|
15: "cuda:1"
|
|
30: "cuda:2"
|
|
45: "cuda:3"
|
|
|
|
- match:
|
|
name: "^model\\.layers\\.([0-9]|[1][0-4])\\."
|
|
replace:
|
|
class: "default"
|
|
kwargs:
|
|
generate_device: "cuda:0"
|
|
prefill_device: "cuda:0"
|
|
- match:
|
|
name: "(^model\\.layers\\.([2][0-9]|[1][5-9])\\.)"
|
|
replace:
|
|
class: "default"
|
|
kwargs:
|
|
generate_device: "cuda:1"
|
|
prefill_device: "cuda:1"
|
|
- match:
|
|
name: "(^model\\.layers\\.([3][0-9]|[4][0-4])\\.)"
|
|
replace:
|
|
class: "default"
|
|
kwargs:
|
|
generate_device: "cuda:2"
|
|
prefill_device: "cuda:2"
|
|
|
|
- match:
|
|
name: "^lm_head"
|
|
class: torch.nn.Linear
|
|
replace:
|
|
class: ktransformers.operators.linear.KTransformersLinear
|
|
kwargs:
|
|
generate_device: "cuda:3"
|
|
prefill_device: "cuda:3"
|
|
generate_op: "KLinearMarlin"
|
|
prefill_op: "KLinearTorch"
|
|
|
|
- match:
|
|
name: "(^model\\.layers\\.([5][0-9]|[4][5-9])\\.)|(^model.norm)"
|
|
replace:
|
|
class: "default"
|
|
kwargs:
|
|
generate_device: "cuda:3"
|
|
prefill_device: "cuda:3" |