vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

change mul_sum_i8_pairs_float to use AVX_VNNI

Browse source
This commit is contained in:
MeouSker77 2023-04-23 11:44:56 +08:00
parent 81cb1eee30
commit d771b8152d
1 changed files with 5 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

10
ggml.c
View file

@ -505,15 +505,15 @@ static inline __m256 sum_i16_pairs_float(const __m256i x) {
// multiply int8_t, add results pairwise twice and return as float vector
static inline __m256 mul_sum_i8_pairs_float(const __m256i x, const __m256i y) {
#if __AVXVNNIINT8__
const __m256i zero = _mm256_setzero_si256();
const __m256i summed_pairs = _mm256_dpbssd_epi32(zero, x, y);
return _mm256_cvtepi32_ps(summed_pairs);
#else
// Get absolute values of x vectors
const __m256i ax = _mm256_sign_epi8(x, x);
// Sign the values of the y vectors
const __m256i sy = _mm256_sign_epi8(y, x);
#if __AVXVNNI__
const __m256i zero = _mm256_setzero_si256();
const __m256i summed_pairs = _mm256_dpbusd_epi32(zero, ax, sy);
return _mm256_cvtepi32_ps(summed_pairs);
#else
// Perform multiplication and create 16-bit values
const __m256i dot = _mm256_maddubs_epi16(ax, sy);
return sum_i16_pairs_float(dot);