Skip to content
This repository has been archived by the owner on Nov 17, 2023. It is now read-only.

Commit

Permalink
[1.8] Include oneDNN gemm fix (#19251)
Browse files Browse the repository at this point in the history 
* Include oneDNN gemm fix

* oneDNN v1.6.4

* Update mkldnn_format_tag_last
  • Loading branch information
@leezu
leezu authored Oct 5, 2020
1 parent f4f820c commit 25bebdf
Show file tree
Hide file tree
Showing 2 changed files with 2 additions and 2 deletions.
2 changes: 1 addition & 1 deletion 3rdparty/mkldnn
Submodule mkldnn updated 53 files
+1 −1 CMakeLists.txt
+34 −24 doc/primitives/matmul.md
+35 −0 include/dnnl.hpp
+14 −0 include/dnnl_types.h
+6 −6 src/CMakeLists.txt
+14 −0 src/common/c_types_map.hpp
+14 −0 src/common/dnnl_debug_autogenerated.cpp
+39 −28 src/common/matmul.cpp
+20 −9 src/common/matmul_pd.hpp
+14 −0 src/common/memory_desc_wrapper.cpp
+14 −0 src/common/tag_traits.hpp
+41 −0 src/common/utils.hpp
+33 −7 src/common/verbose.cpp
+44 −5 src/cpu/gemm_convolution_utils.cpp
+1 −1 src/cpu/gemm_inner_product.cpp
+43 −15 src/cpu/gemm_inner_product_utils.cpp
+10 −5 src/cpu/gemm_inner_product_utils.hpp
+1 −1 src/cpu/gemm_x8s8s32x_inner_product.cpp
+32 −3 src/cpu/matmul/gemm_based_common.hpp
+42 −49 src/cpu/matmul/gemm_bf16_matmul.cpp
+2 −2 src/cpu/matmul/gemm_bf16_matmul.hpp
+40 −49 src/cpu/matmul/gemm_f32_matmul.cpp
+2 −2 src/cpu/matmul/gemm_f32_matmul.hpp
+52 −54 src/cpu/matmul/gemm_x8s8s32x_matmul.cpp
+2 −2 src/cpu/matmul/gemm_x8s8s32x_matmul.hpp
+78 −0 src/cpu/matmul/matmul_utils.hpp
+44 −31 src/cpu/matmul/ref_matmul.cpp
+7 −5 src/cpu/ref_convolution.cpp
+10 −7 src/cpu/rnn/ref_rnn.cpp
+3 −3 src/cpu/x64/gemm/f32/jit_avx512_core_gemm_smalln_tn_f32_kern.cpp
+29 −15 src/cpu/x64/gemm/gemv_driver.cpp
+1 −1 src/cpu/x64/gemm_bf16_inner_product.cpp
+17 −6 src/cpu/x64/jit_avx2_1x1_conv_kernel_f32.cpp
+58 −16 src/cpu/x64/jit_gemm_inner_product_utils.cpp
+2 −2 src/cpu/x64/jit_gemm_inner_product_utils.hpp
+2 −1 src/cpu/x64/jit_uni_1x1_conv_utils.hpp
+6 −0 tests/benchdnn/dnn_types.cpp
+14 −0 tests/benchdnn/dnnl_debug_autogenerated.cpp
+7 −3 tests/benchdnn/dnnl_memory.hpp
+32 −6 tests/benchdnn/doc/driver_matmul.md
+2 −2 tests/benchdnn/inputs/bnorm/set_topologies_gpu
+1 −0 tests/benchdnn/inputs/matmul/shapes_3d
+20 −0 tests/benchdnn/inputs/matmul/shapes_multidim
+4 −1 tests/benchdnn/inputs/matmul/test_matmul_all
+14 −0 tests/benchdnn/inputs/matmul/test_matmul_multidims
+14 −2 tests/benchdnn/matmul/bench_matmul.cpp
+114 −51 tests/benchdnn/matmul/matmul.cpp
+121 −28 tests/benchdnn/matmul/matmul.hpp
+95 −21 tests/benchdnn/matmul/matmul_aux.cpp
+16 −7 tests/benchdnn/matmul/ref_matmul.cpp
+11 −3 tests/gtests/test_ip_formats.cpp
+20 −0 tests/gtests/test_matmul.cpp
+8 −1 tests/gtests/test_reorder_formats.cpp
2 changes: 1 addition & 1 deletion tests/cpp/operator/mkldnn_test.cc
View file
Original file line number Diff line number Diff line change
Expand Up @@ -100,7 +100,7 @@ static void VerifyDefMem(const mkldnn::memory &mem) {

TEST(MKLDNN_UTIL_FUNC, MemFormat) {
// Check whether the number of format is correct.
CHECK_EQ(mkldnn_format_tag_last, 205);
CHECK_EQ(mkldnn_format_tag_last, 219);
CHECK_EQ(mkldnn_nchw, 5);
CHECK_EQ(mkldnn_oihw, 5);
}
Expand Down

0 comments on commit 25bebdf

Please sign in to comment.