tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-26 19:55:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

ggml-cpu: chore: remove todo comments about inline asm

Browse Source 
Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>
This commit is contained in:
Aaron Teo
2025-06-20 21:49:41 +08:00
parent 70ff4e6b31
commit 2b4892e20f
1 changed files with 0 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
ggml/src/ggml-impl.h
View File

@ -426,14 +426,12 @@ GGML_API void ggml_aligned_free(void * ptr, size_t size);
#define GGML_FP16_TO_FP32(x) GGML_COMPUTE_FP16_TO_FP32(x)
#define GGML_FP32_TO_FP16(x) GGML_COMPUTE_FP32_TO_FP16(x)
// TODO: Determine if inline assembly is faster
static inline float ggml_compute_fp16_to_fp32(ggml_fp16_t h) {
uint16x8_t v_h = vec_splats(h);
uint16x8_t nnpa_dlf16 = vec_convert_from_fp16(v_h, 0);
return vec_extend_to_fp32_hi(nnpa_dlf16, 0)[0];
}
// TODO: Determine if inline assembly is faster
static inline ggml_fp16_t ggml_compute_fp32_to_fp16(float f) {
float32x4_t v_f = vec_splats(f);
float32x4_t v_zero = vec_splats(0.0f);