diff options
| author | junchao-loongson <68935141+junchao-loongson@users.noreply.github.com> | 2024-05-20 15:19:21 +0800 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-05-20 10:19:21 +0300 |
| commit | 65c58207ece92ad213f4bfd0f91dcb2dfb664f5b (patch) | |
| tree | b6fcdb6e1ee912b34e0ce592edc60d71b6cbbc1d /ggml-impl.h | |
| parent | 1cc0155d04918cb3017afa472acea51b77483c4a (diff) | |
ggml : add loongarch lsx and lasx support (#6454)
* add loongarch lsx and lasx optimize code
* Add loongarch compilation support to makefile
* revert stb_image.h
* opt bytes_from_nibbles_32 and sum_i16_pairs_float
* fix undeclared
* format code
* update
* update 2
---------
Co-authored-by: Jinyang He <hejinyang@loongson.cn>
Diffstat (limited to 'ggml-impl.h')
| -rw-r--r-- | ggml-impl.h | 28 |
1 files changed, 28 insertions, 0 deletions
diff --git a/ggml-impl.h b/ggml-impl.h index 5ff014fe..362d40f4 100644 --- a/ggml-impl.h +++ b/ggml-impl.h @@ -455,6 +455,34 @@ static inline ggml_fp16_t ggml_compute_fp32_to_fp16(float f) { #include <riscv_vector.h> #endif +#if defined(__loongarch64) +#if defined(__loongarch_asx) +#include <lasxintrin.h> +#endif +#if defined(__loongarch_sx) +#include <lsxintrin.h> +#endif +#endif + +#if defined(__loongarch_asx) + +typedef union { + int32_t i; + float f; +} ft_union; + +/* float type data load instructions */ +static __m128 __lsx_vreplfr2vr_s(float val) { + ft_union fi_tmpval = {.f = val}; + return (__m128)__lsx_vreplgr2vr_w(fi_tmpval.i); +} + +static __m256 __lasx_xvreplfr2vr_s(float val) { + ft_union fi_tmpval = {.f = val}; + return (__m256)__lasx_xvreplgr2vr_w(fi_tmpval.i); +} +#endif + #ifdef __F16C__ #ifdef _MSC_VER |