Skip to content

[FORK][FIX] Fix performance issues caused by using the thread pool #291

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

Closed
wants to merge 1 commit into from
Closed

[FORK][FIX] Fix performance issues caused by using the thread pool #291

Show file tree
Hide file tree
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
2 changes: 1 addition & 1 deletion src/cpu/matmul/cpu_matmul_list.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -95,11 +95,11 @@ const impl_list_item_t impl_list[] = REG_MATMUL_P({
CPU_INSTANCE_AVX512(brgemm_matmul_t,avx512_core)
CPU_INSTANCE_AVX2(brgemm_matmul_t,avx2_vnni_2)
CPU_INSTANCE_AVX2(brgemm_matmul_t,avx2_vnni)
CPU_INSTANCE_AVX2(brgemm_matmul_t, avx2)
CPU_INSTANCE(gemm_f32_matmul_t)
CPU_INSTANCE(gemm_bf16_matmul_t, f32)
CPU_INSTANCE(gemm_bf16_matmul_t, bf16)
CPU_INSTANCE(gemm_x8s8s32x_matmul_t)
CPU_INSTANCE_AVX2(brgemm_matmul_t, avx2)
CPU_INSTANCE(ref_matmul_t)
CPU_INSTANCE(ref_matmul_int8_t)
// These implementations are enabled only when DNNL_EXPERIMENTAL_SPARSE
Expand Down
8 changes: 0 additions & 8 deletions src/cpu/x64/jit_brgemm_conv_utils.cpp
View file
Open in desktop
Copy link
Collaborator

@maxnick maxnick Sep 23, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think it does make sense to simply remove this code, rather than keep it commented.

Copy link
Collaborator Author

@sunxiaoxia2022 sunxiaoxia2022 Sep 29, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok, updated.

Original file line number Diff line number Diff line change
Expand Up @@ -2296,10 +2296,6 @@ status_t init_conf(jit_brgemm_conv_conf_t &jcp, cpu_isa_t isa,

if (try_exec_type_res == false) return status::unimplemented;

#if DNNL_CPU_THREADING_RUNTIME == DNNL_RUNTIME_THREADPOOL
adjust_nthr(jcp, src_d, dst_d);
#endif

// ============ end blocking ===========================================
jcp.brg_type
= (jcp.use_uker && one_of(jcp.exec_type, exec_base, exec_trans))
Expand Down Expand Up @@ -2530,10 +2526,6 @@ status_t init_1x1_conf(jit_brgemm_conv_conf_t &jcp, cpu_isa_t isa,
}
best_brgb.save_to_jcp(jcp);

#if DNNL_CPU_THREADING_RUNTIME == DNNL_RUNTIME_THREADPOOL
adjust_nthr(jcp, src_d, dst_d);
#endif

// =============== end blocking =================================

jcp.brg_stride_a = jcp.ic_block * jcp.src_dsz;
Expand Down
Loading