CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-28 08:31:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

ggml : support AVX512VNNI (#6280)

Browse Source 
This change causes some quants (e.g. Q4_0, Q8_0) to go faster on some
architectures (e.g. AMD Zen 4).
This commit is contained in:
Justine Tunney
2024-03-25 01:39:56 -04:00
committed by GitHub
parent a32b77c4b2
commit 7733f0c760
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
ggml-quants.c
View File

@@ -132,7 +132,7 @@ static inline __m256 sum_i16_pairs_float(const __m256i x) {
}
static inline __m256 mul_sum_us8_pairs_float(const __m256i ax, const __m256i sy) {
#if __AVXVNNI__
#if defined(__AVXVNNI__) || defined(__AVX512VNNI__)
const __m256i zero = _mm256_setzero_si256();
const __m256i summed_pairs = _mm256_dpbusd_epi32(zero, ax, sy);
return _mm256_cvtepi32_ps(summed_pairs);