happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

ggml : arm repack fix build (whisper/0)

This commit is contained in:
Georgi Gerganov 2025-12-13 08:04:09 +02:00
parent 615655aafe
commit 71fdcf0616
1 changed files with 2 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
ggml/src/ggml-cpu/arch/arm/repack.cpp
View File

@ -24,6 +24,7 @@
#define UNUSED GGML_UNUSED #define UNUSED GGML_UNUSED
#if defined(__aarch64__) && defined(__ARM_NEON) && defined(__ARM_FEATURE_MATMUL_INT8)
static inline void decode_q4_Kx8_scales_mins(const uint8_t * scales_in, static inline void decode_q4_Kx8_scales_mins(const uint8_t * scales_in,
int16x8_t * out_mins, int16x8_t * out_mins,
int8_t * out_scales) { int8_t * out_scales) {
@ -46,6 +47,7 @@ static inline void decode_q4_Kx8_scales_mins(const uint8_t * scales_in,
scales_u32[1] = (sm[2] & kmask2) | (((sm[0] >> 6) & kmask3) << 4); scales_u32[1] = (sm[2] & kmask2) | (((sm[0] >> 6) & kmask3) << 4);
memcpy(out_scales, scales_u32, 8); memcpy(out_scales, scales_u32, 8);
} }
#endif
void ggml_quantize_mat_q8_0_4x4(const float * GGML_RESTRICT x, void * GGML_RESTRICT vy, int64_t k) { void ggml_quantize_mat_q8_0_4x4(const float * GGML_RESTRICT x, void * GGML_RESTRICT vy, int64_t k) {
assert(QK8_0 == 32); assert(QK8_0 == 32);