diff options
| author | Justine Tunney <jtunney@gmail.com> | 2024-03-25 01:39:56 -0400 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-03-25 07:39:56 +0200 |
| commit | 7733f0c76081b2a69b5f8b192db2db7c43629d58 (patch) | |
| tree | 2a78e3e47fbd4d768d61f46d06c5c2815640595e | |
| parent | a32b77c4b2c1808654d0b952f26c37d73d2e746b (diff) | |
ggml : support AVX512VNNI (#6280)
This change causes some quants (e.g. Q4_0, Q8_0) to go faster on some
architectures (e.g. AMD Zen 4).
| -rw-r--r-- | ggml-quants.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/ggml-quants.c b/ggml-quants.c index 2eaca059..f26798ac 100644 --- a/ggml-quants.c +++ b/ggml-quants.c @@ -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); |