diff options
| author | Kawrakow <iwankawrakow@gmail.com> | 2025-06-11 13:05:26 +0300 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-06-11 13:05:26 +0300 |
| commit | 69af3f5990327df4f8dcb05817825802b7e6bed8 (patch) | |
| tree | 1ddd396489b53e0f20b12d6701451af8694674f8 /ggml/src/ggml.c | |
| parent | e56061fa12e5fcb7a9a8fa5fca9e74c20d166bcc (diff) | |
Much faster iq3_xxs GEMM via repacking to q8_0_r8 (AVX2) (#516)
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'ggml/src/ggml.c')
| -rw-r--r-- | ggml/src/ggml.c | 4 |
1 files changed, 4 insertions, 0 deletions
diff --git a/ggml/src/ggml.c b/ggml/src/ggml.c index 9e3c4b90..bf5a8519 100644 --- a/ggml/src/ggml.c +++ b/ggml/src/ggml.c @@ -1123,7 +1123,11 @@ static const ggml_type_traits_t type_traits[GGML_TYPE_COUNT] = { .from_float = quantize_row_iq3_xxs, .from_float_ref = (ggml_from_float_t)quantize_row_iq3_xxs_ref, .vec_dot = ggml_vec_dot_iq3_xxs_q8_K, +#ifdef __AVX2__ + .vec_dot_type = GGML_TYPE_Q8_2_X4, +#else .vec_dot_type = GGML_TYPE_Q8_K, +#endif .nrows = 1, .row_meta_size = 0, }, |