EngineX-Ascend/enginex-ascend-910-llama.cpp
10
0
Fork 0
Code Issues Pull Requests Actions 4 Projects Releases Wiki Activity

metal : improve F32, F16 and BF16 mat-vec multiplication (#16057)

Browse Source 
* metal : improve F32, F16 and BF16 mat-vec multiplication

ggml-ci

* metal : make the NSG a function constant in mul_mv kernels

ggml-ci
This commit is contained in:
Georgi Gerganov
2025-09-18 12:33:45 +03:00
committed by GitHub
parent e00f3fd8ff
commit b213fce89b
6 changed files with 355 additions and 288 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
ggml/src/ggml-metal/ggml-metal-device.h
View File

@@ -22,6 +22,7 @@ typedef struct ggml_metal_cv * ggml_metal_cv_t;
ggml_metal_cv_t ggml_metal_cv_init(void);
void ggml_metal_cv_free(ggml_metal_cv_t cv);
void ggml_metal_cv_set_int16(ggml_metal_cv_t cv, int16_t value, int32_t idx);
void ggml_metal_cv_set_int32(ggml_metal_cv_t cv, int32_t value, int32_t idx);
void ggml_metal_cv_set_bool (ggml_metal_cv_t cv, bool value, int32_t idx);