vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 01:24:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'upstream' into concedo_experimental

Browse source 
# Conflicts:
#	.github/workflows/build.yml
#	ggml/cmake/ggml-config.cmake.in
#	ggml/src/ggml-cann/CMakeLists.txt
#	ggml/src/ggml-cann/common.h
#	ggml/src/ggml-cann/ggml-cann.cpp
#	ggml/src/ggml-cuda/fattn.cu
#	ggml/src/ggml-opencl/CMakeLists.txt
#	ggml/src/ggml-opencl/ggml-opencl.cpp
#	requirements/requirements-convert_hf_to_gguf.txt
#	scripts/compare-llama-bench.py
#	tests/test-chat-template.cpp
#	tests/test-chat.cpp
#	tools/llama-bench/llama-bench.cpp
This commit is contained in:
Concedo 2025-08-07 21:23:09 +08:00
commit 8a71eb03c0
29 changed files with 1350 additions and 291 deletions
Download patch file Download diff file Expand all files Collapse all files

9
ggml/src/ggml-backend.cpp
View file

@ -1077,6 +1077,11 @@ static void ggml_backend_sched_split_graph(ggml_backend_sched_t sched, struct gg
}
}
}
// if the node is still unassigned, assign it to the first backend that supports it
for (int b = 0; b < sched->n_backends && *cur_backend_id == -1; b++) {
ggml_backend_sched_set_if_supported(sched, node, b, cur_backend_id);
}
GGML_ASSERT(*cur_backend_id != -1);
}
// pass 5: split graph, find tensors that need to be copied
@ -1104,7 +1109,7 @@ static void ggml_backend_sched_split_graph(ggml_backend_sched_t sched, struct gg
const int node_backend_id = tensor_backend_id(node);
assert(node_backend_id != -1); // all nodes should be assigned by now, this can happen if there is no CPU fallback
GGML_ASSERT(node_backend_id != -1); // all nodes should be assigned by now, this can happen if there is no CPU fallback
// check if we should start a new split based on the sources of the current node
bool need_new_split = false;
@ -1162,7 +1167,7 @@ static void ggml_backend_sched_split_graph(ggml_backend_sched_t sched, struct gg
size_t src_id = hash_id(src);
const int src_backend_id = sched->hv_tensor_backend_ids[src_id];
assert(src_backend_id != -1); // all inputs should be assigned by now
GGML_ASSERT(src_backend_id != -1); // all inputs should be assigned by now
if (src->flags & GGML_TENSOR_FLAG_INPUT && sched->n_copies > 1) {
if (tensor_id_copy(src_id, src_backend_id, 0) == NULL) {