Skip to content

Commit

Permalink
Revert "[SYCL] fallback mmvq (ggerganov#9088)" (ggerganov#9579)
Browse files Browse the repository at this point in the history
This reverts commit 50addec.
  • Loading branch information
Akarshan Biswas authored Sep 23, 2024
1 parent c35e586 commit e62e978
Show file tree
Hide file tree
Showing 2 changed files with 1 addition and 3 deletions.
3 changes: 1 addition & 2 deletions ggml/src/ggml-sycl.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -3496,8 +3496,7 @@ static void ggml_sycl_mul_mat(ggml_backend_sycl_context & ctx, const ggml_tensor

bool use_mul_mat_vec_q = ggml_is_quantized(src0->type)
&& src1->type == GGML_TYPE_F32 && dst->type == GGML_TYPE_F32
&& src1->ne[1] <= MMVQ_MAX_BATCH_SIZE
&& (ctx.stream()->get_backend() == sycl::backend::ext_oneapi_cuda || src1->ne[1] > MMVQ_MIN_BATCH_SIZE);
&& src1->ne[1] <= MMVQ_MAX_BATCH_SIZE;

bool use_mul_mat_q = ggml_sycl_supports_mmq(src0->type)
&& src1->type == GGML_TYPE_F32 && dst->type == GGML_TYPE_F32;
Expand Down
1 change: 0 additions & 1 deletion ggml/src/ggml-sycl/common.hpp
Original file line number Diff line number Diff line change
Expand Up @@ -134,7 +134,6 @@ typedef sycl::float2 dfloat2;
#endif // GGML_SYCL_F16

#define MMVQ_MAX_BATCH_SIZE 8
#define MMVQ_MIN_BATCH_SIZE 4

static const int8_t kvalues_iq4nl[16]={-127, -104, -83, -65, -49, -35, -22, -10, 1, 13, 25, 38, 53, 69, 89, 113};

Expand Down

0 comments on commit e62e978

Please sign in to comment.