Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

metal : consolidate bin kernels (#19390)

Browse Source 
* metal : refactor bin kernels

* cont

* cont : fix cv
This commit is contained in:
Georgi Gerganov
2026-02-07 10:35:56 +02:00
committed by GitHub
parent 34ba7b5a2f
commit 8872ad2125
6 changed files with 224 additions and 295 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-metal/ggml-metal-device.h
+5 -1
View File
@@ -53,6 +53,9 @@ struct ggml_metal_pipeline_with_params {
int nr1;
size_t smem;
bool c4;
bool cnt;
};
int ggml_metal_pipeline_max_theads_per_threadgroup(struct ggml_metal_pipeline_with_params pipeline);
@@ -134,7 +137,8 @@ struct ggml_metal_pipeline_with_params ggml_metal_library_get_pipeline_argsort
struct ggml_metal_pipeline_with_params ggml_metal_library_get_pipeline_argsort_merge (ggml_metal_library_t lib, const struct ggml_tensor * op);
struct ggml_metal_pipeline_with_params ggml_metal_library_get_pipeline_top_k (ggml_metal_library_t lib, const struct ggml_tensor * op);
struct ggml_metal_pipeline_with_params ggml_metal_library_get_pipeline_top_k_merge (ggml_metal_library_t lib, const struct ggml_tensor * op);
struct ggml_metal_pipeline_with_params ggml_metal_library_get_pipeline_bin (ggml_metal_library_t lib, enum ggml_op op, int32_t n_fuse, bool row);
struct ggml_metal_pipeline_with_params ggml_metal_library_get_pipeline_bin (ggml_metal_library_t lib, const struct ggml_tensor * op, int32_t n_fuse );
struct ggml_metal_pipeline_with_params ggml_metal_library_get_pipeline_bin_one (ggml_metal_library_t lib, enum ggml_op op);
struct ggml_metal_pipeline_with_params ggml_metal_library_get_pipeline_l2_norm (ggml_metal_library_t lib, const struct ggml_tensor * op);
struct ggml_metal_pipeline_with_params ggml_metal_library_get_pipeline_group_norm (ggml_metal_library_t lib, const struct ggml_tensor * op);
struct ggml_metal_pipeline_with_params ggml_metal_library_get_pipeline_norm (ggml_metal_library_t lib, const struct ggml_tensor * op, int32_t n_fuse);