Skip to content

sycl: route small f32 matmuls to oneMKL, bypass oneDNN #22150

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
aicss-genai wants to merge 2 commits into
base: master
Choose a base branch
from
+17 −13
Open

sycl: route small f32 matmuls to oneMKL, bypass oneDNN #22150

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1dd517d
sycl: route small f32 matmuls to oneMKL, bypass oneDNN
ctao456 Apr 17, 2026
47d6ee8
Merge branch 'ggml-org:master' into aicss-genai/sycl-bmg-upstream-pr-4
aicss-genai Apr 20, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
30 changes: 17 additions & 13 deletions ggml/src/ggml-sycl/ggml-sycl.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2253,21 +2253,25 @@ inline void ggml_sycl_op_mul_mat_sycl(
const float * src0_ddf_i = src0->type == GGML_TYPE_F32 ? (const float *) src0_dd_i : src0_ddq_as_f32.get();
const float * src1_ddf1_i = src1->type == GGML_TYPE_F32 ? (const float *) src1_ddf_i : src1_ddq_as_f32.get();

{
const int64_t gemm_flops = (int64_t)row_diff * src1_ncols * ne10;
const bool use_mkl_direct = gemm_flops < 256 * 256 * 256;
#if GGML_SYCL_DNNL
if (!g_ggml_sycl_disable_dnn) {
DnnlGemmWrapper::row_gemm(ctx, row_diff, src1_ncols, ne10, src0_ddf_i,
DnnlGemmWrapper::to_dt<float>(), src1_ddf1_i, DnnlGemmWrapper::to_dt<float>(),
dst_dd_i, DnnlGemmWrapper::to_dt<float>(), stream);
}
else
if (!g_ggml_sycl_disable_dnn && !use_mkl_direct) {
DnnlGemmWrapper::row_gemm(ctx, row_diff, src1_ncols, ne10, src0_ddf_i,
DnnlGemmWrapper::to_dt<float>(), src1_ddf1_i, DnnlGemmWrapper::to_dt<float>(),
dst_dd_i, DnnlGemmWrapper::to_dt<float>(), stream);
}
else
#endif
{
const float alpha = 1.0f;
const float beta = 0.0f;
SYCL_CHECK(CHECK_TRY_ERROR(oneapi::mkl::blas::column_major::gemm(
*stream, oneapi::mkl::transpose::trans, oneapi::mkl::transpose::nontrans, row_diff,
src1_ncols, ne10, dpct::get_value(&alpha, *stream), src0_ddf_i, ne00, src1_ddf1_i, ne10,
dpct::get_value(&beta, *stream), dst_dd_i, ldc)));
{
const float alpha = 1.0f;
const float beta = 0.0f;
SYCL_CHECK(CHECK_TRY_ERROR(oneapi::mkl::blas::column_major::gemm(
*stream, oneapi::mkl::transpose::trans, oneapi::mkl::transpose::nontrans, row_diff,
src1_ncols, ne10, dpct::get_value(&alpha, *stream), src0_ddf_i, ne00, src1_ddf1_i, ne10,
dpct::get_value(&beta, *stream), dst_dd_i, ldc)));
}
}
}
GGML_UNUSED(dst);
Expand Down