Skip to content

Commit

Permalink
[GPU] Update oneDNN v3.5 (openvinotoolkit#24663)
Browse files Browse the repository at this point in the history 
Applied oneDNN convolution improvement in Stable-Diffusion LNL.

### Tickets:
 - *ticket-id*

Signed-off-by: hyunback <hyunback.kim@intel.com>
  • Loading branch information
@hyunback
hyunback authored May 24, 2024
1 parent 0297fac commit 8682e5e
Showing 1 changed file with 1 addition and 1 deletion.
2 changes: 1 addition & 1 deletion src/plugins/intel_gpu/thirdparty/onednn_gpu
Submodule onednn_gpu updated 62 files
+1 −1 doc/Doxyfile.in
+13 −12 doc/build/link.md
+1 −1 doc/graph/operations/Dequantize.md
+1 −2 doc/graph/operations/DynamicDequantize.md
+4 −5 doc/graph/operations/DynamicQuantize.md
+1 −1 doc/graph/operations/Quantize.md
+2 −1 doc/performance_considerations/verbose.md
+98 −0 doc/performance_considerations/verbose_table.md
+1 −0 doc/rst/index.rst
+1 −0 doc/rst/performance_profiling_and_inspection.rst
+10 −0 doc/rst/ukernels.rst
+1 −1 doc/sphinx/conf.py
+25 −37 doc/ukernel/operations/brgemm.md
+5 −5 doc/ukernel/operations/transform.md
+1 −1 examples/CMakeLists.txt
+33 −19 examples/ukernels/cpu_brgemm.cpp
+3 −3 include/oneapi/dnnl/dnnl_ukernel.h
+5 −3 include/oneapi/dnnl/dnnl_ukernel.hpp
+1 −1 src/common/verbose_msg.hpp
+9 −5 src/cpu/aarch64/acl_deconvolution.hpp
+1 −6 src/gpu/intel/compute/compute_engine.hpp
+10 −16 src/gpu/intel/compute/device_info.cpp
+4 −2 src/gpu/intel/compute/device_info.hpp
+7 −0 src/gpu/intel/compute/kernel_arg_list.hpp
+193 −2 src/gpu/intel/jit/codegen/kernel.hpp
+389 −40 src/gpu/intel/jit/conv/config.cpp
+31 −5 src/gpu/intel/jit/conv/config.hpp
+3 −2 src/gpu/intel/jit/conv/conv_kernel.hpp
+6 −0 src/gpu/intel/jit/conv/ir_builder.cpp
+752 −869 src/gpu/intel/jit/conv/lookup_table_data.cpp
+68 −97 src/gpu/intel/jit/conv/plan.cpp
+1 −0 src/gpu/intel/jit/conv/tiler.cpp
+19 −6 src/gpu/intel/jit/gemm/gen_gemm.hpp
+20 −0 src/gpu/intel/jit/gemm/gen_gemm_kernel.hpp
+112 −106 src/gpu/intel/jit/gemm/gen_gemm_kernel_generator.cpp
+17 −4 src/gpu/intel/jit/gemm/gen_gemm_kernel_generator.hpp
+771 −686 src/gpu/intel/jit/gemm/kernel.db
+1 −1 src/gpu/intel/jit/gemm/strategy_parser.cpp
+12 −0 src/gpu/intel/jit/gemm/utils.hpp
+20 −15 src/gpu/intel/jit/ir/config.hpp
+12 −0 src/gpu/intel/jit/ir/core.hpp
+33 −2 src/gpu/intel/jit/ir/gemm_schedule.hpp
+5 −0 src/gpu/intel/jit/ir/hw.hpp
+5 −17 src/gpu/intel/jit/ir/ir.cpp
+9 −0 src/gpu/intel/jit/ir/problem.cpp
+2 −1 src/gpu/intel/jit/ir/problem.hpp
+186 −0 src/gpu/intel/jit/ir/walk_order.hpp
+106 −206 src/gpu/intel/jit/pass/cse.cpp
+3 −3 src/gpu/intel/jit/utils/utils.hpp
+6 −0 src/gpu/intel/ocl/ocl_gpu_device_info.cpp
+10 −0 src/gpu/intel/ocl/ref_matmul.hpp
+1 −0 src/gpu/sycl/sycl_interop_gpu_kernel.cpp
+6 −4 src/graph/backend/dnnl/kernels/mqa.hpp
+4 −1 src/graph/backend/dnnl/kernels/sdp.hpp
+2 −0 src/graph/backend/dnnl/platform.cpp
+2 −0 src/sycl/sycl_device_info.cpp
+1 −0 tests/benchdnn/inputs/deconv/shapes_1d
+6 −6 tests/benchdnn/inputs/matmul/test_matmul_ci
+4 −4 tests/benchdnn/inputs/matmul/test_matmul_gpu
+4 −0 tests/benchdnn/matmul/cfg.cpp
+7 −1 tests/benchdnn/matmul/matmul.cpp
+4 −2 tests/gtests/graph/unit/utils.hpp

0 comments on commit 8682e5e

Please sign in to comment.