diff options
| author | AidanBeltonS <87009434+AidanBeltonS@users.noreply.github.com> | 2024-05-27 17:34:51 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-05-27 22:04:51 +0530 |
| commit | 95f84d5ce8b449a9b16009434aca800df504a02e (patch) | |
| tree | 6031df988d57149b4b31c9cdcc0bffc0cc92f074 | |
| parent | 5487593bc7ee0b65b9d2e2985b4b61dc77043101 (diff) | |
Fix q_xxs using mul_mat_q (#7459)
| -rw-r--r-- | ggml-sycl.cpp | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/ggml-sycl.cpp b/ggml-sycl.cpp index f329bc27..8839f775 100644 --- a/ggml-sycl.cpp +++ b/ggml-sycl.cpp @@ -15263,6 +15263,7 @@ static void ggml_sycl_mul_mat(const ggml_tensor * src0, const ggml_tensor * src1 } } else { bool use_mul_mat_q = min_compute_capability >= VER_4VEC && ggml_is_quantized(src0->type); + use_mul_mat_q = use_mul_mat_q && (src0->type != GGML_TYPE_IQ2_XXS); if (use_xmx && min_compute_capability >= VER_GEN9 && src1->ne[1] > XMX_MAX_BATCH_SIZE) { use_mul_mat_q = false; |