Skip to content
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

[SYCL]rm wait() to improve the performance #7233

Merged
merged 1 commit into from
May 13, 2024
Merged

[SYCL]rm wait() to improve the performance #7233

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
25 changes: 1 addition & 24 deletions ggml-sycl.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15656,26 +15656,6 @@ static void ggml_sycl_mul_mat_batched_sycl(const ggml_tensor *src0,
const int64_t r2 = ne12/ne02;
const int64_t r3 = ne13/ne03;

#if 0
// use syclGemmEx
{
for (int i13 = 0; i13 < ne13; ++i13) {
for (int i12 = 0; i12 < ne12; ++i12) {
int i03 = i13 / r3;
int i02 = i12 / r2;

SYCL_CHECK(
syclGemmEx(g_sycl_handles[g_main_device], CUBLAS_OP_T, CUBLAS_OP_N,
ne01, ne11, ne10,
alpha, (const char *) src0_as_f16 + i02*src0->nb[2] + i03*src0->nb[3] , SYCL_R_16F, nb01/sizeof(half),
(const char *) src1_as_f16 + i12*src1->nb[2]/2 + i13*src1->nb[3]/2, SYCL_R_16F, nb11/sizeof(float),
beta, ( char *) dst_t + i12*nbd2 + i13*nbd3, cu_data_type, ne01,
cu_compute_type,
CUBLAS_GEMM_DEFAULT_TENSOR_OP));
}
}
}
#else
if (r2 == 1 && r3 == 1 && src0->nb[2]*src0->ne[2] == src0->nb[3] && src1->nb[2]*src1->ne[2] == src1->nb[3]) {
// there is no broadcast and src0, src1 are contiguous across dims 2, 3
SYCL_CHECK(CHECK_TRY_ERROR(dpct::gemm_batch(
Expand All @@ -15687,7 +15667,6 @@ static void ggml_sycl_mul_mat_batched_sycl(const ggml_tensor *src0,
nb11 / nb10, nb12 / nb10, beta,
(char *)dst_t, cu_data_type, ne01, nb2 / nb0,
ne12 * ne13, cu_compute_type)));
g_sycl_handles[g_main_device]->wait();
} else {
const int ne23 = ne12*ne13;

Expand Down Expand Up @@ -15718,7 +15697,7 @@ static void ggml_sycl_mul_mat_batched_sycl(const ggml_tensor *src0,
nb02, nb03, nb12_scaled, nb13_scaled,
nbd2, nbd3, r2, r3, item_ct1);
});
}).wait();
});
}
SYCL_CHECK(CHECK_TRY_ERROR(dpct::gemm_batch(
*g_sycl_handles[g_main_device], oneapi::mkl::transpose::trans,
Expand All @@ -15729,9 +15708,7 @@ static void ggml_sycl_mul_mat_batched_sycl(const ggml_tensor *src0,
dpct::library_data_t::real_half, nb11 / nb10, beta,
(void **)(ptrs_dst.get() + 0 * ne23), cu_data_type, ne01, ne23,
cu_compute_type)));
g_sycl_handles[g_main_device]->wait();
}
#endif

if (no_mixed_dtypes) {
const to_fp32_sycl_t to_fp32_sycl = ggml_get_to_fp32_sycl(GGML_TYPE_F16);
Expand Down
Loading