MIOpen 6.2.0 file not found grouped_convolution_forward_bilinear.hpp

[RandUser123sa]

Hello,

I'm trying to build MIOpen 6.2.0 with params bellow:

export CC=/opt/rocm/llvm/bin/clang
export CXX=/opt/rocm/llvm/bin/clang++

CXX=$ROCM_INSTALL_DIR/llvm/bin/clang++ cmake \
    -Wno-dev \
    -G Ninja \
    -D CMAKE_CXX_FLAGS="${CXXFLAGS} -fcf-protection=none -DNDEBUG" \
    -D CMAKE_INSTALL_PREFIX=/opt/rocm \
    -D CMAKE_BUILD_TYPE=Release \
    -D MIOPEN_BACKEND=HIP \
    -D HALF_INCLUDE_DIR=/usr/include/half \
    -D BUILD_TESTING=NO \
    -D Boost_USE_STATIC_LIBS=NO \
    ..

and receive error on compilation process.

[287/554] Building CXX object src/CMakeFiles/MIOpen.dir/solver/conv_hip_implicit_gemm_3d_grouped_fwd_xdlops.cpp.o
FAILED: src/CMakeFiles/MIOpen.dir/solver/conv_hip_implicit_gemm_3d_grouped_fwd_xdlops.cpp.o 
/opt/rocm/llvm/bin/clang++ -DBOOST_ALL_NO_LIB=1 -DBOOST_ATOMIC_DYN_LINK -DBOOST_ATOMIC_NO_LIB -DBOOST_FILESYSTEM_DYN_LINK -DBOOST_FILESYSTEM_NO_LIB -DHIP_COMPILER_FLAGS=" -x hip   -D__HIP_PLATFORM_AMD__=1  -DUSE_PROF_API=1 --hip-link   -fno-offload-uniform-block " -DMIOPEN_BETA_API=1 -DMIOpen_EXPORTS -DUSE_PROF_API=1 -D__HIP_PLATFORM_AMD__=1 -I/mnt/arhiv/rocm/release/MIOpen-rocm-6.2.0/src/include -I/mnt/arhiv/rocm/rocm-build/build/miopen/include -I/mnt/arhiv/rocm/release/MIOpen-rocm-6.2.0/include -isystem /usr/include/half -isystem /opt/rocm/include/ck -isystem /opt/rocm/include/ck/utility -isystem /opt/rocm/include/ck/tensor_description -isystem /opt/rocm/include/ck/tensor -isystem /opt/rocm/include/ck/problem_transform -isystem /opt/rocm/include/ck/tensor_operation/gpu/device -isystem /opt/rocm/include/ck/tensor_operation/gpu/device/impl -isystem /opt/rocm/include/ck/tensor_operation/gpu/grid -isystem /opt/rocm/include/ck/tensor_operation/gpu/block -isystem /opt/rocm/include/ck/tensor_operation/gpu/warp -isystem /opt/rocm/include/ck/tensor_operation/gpu/thread -isystem /opt/rocm/include/ck/tensor_operation/gpu/element -isystem /opt/rocm/include/ck/library/utility -isystem /opt/rocm/include/ck/library/tensor_operation_instance -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/quantization -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/softmax -isystem /opt/rocm/include -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/conv_tensor_rearrange -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_data -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_weight -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/contraction -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/reduce -isystem /usr/include/eigen3 -isystem /opt/rocm/include/rocmlir -isystem /opt/rocm/include/rocmlir/external/include -fcf-protection=none -DNDEBUG -O3 -DNDEBUG -s -std=c++17 -fPIC -fvisibility-inlines-hidden   -U__HCC__ -Wall -Wextra -Wcomment -Wendif-labels -Wformat -Winit-self -Wreturn-type -Wsequence-point -Wswitch -Wtrigraphs -Wundef -Wuninitialized -Wunreachable-code -Wunused -Wno-ignored-qualifiers -Wno-sign-compare -Weverything -Wno-c++98-compat -Wno-c++98-compat-pedantic -Wno-conversion -Wno-double-promotion -Wno-exit-time-destructors -Wno-extra-semi -Wno-extra-semi-stmt -Wno-float-conversion -Wno-gnu-anonymous-struct -Wno-gnu-zero-variadic-macro-arguments -Wno-missing-prototypes -Wno-nested-anon-types -Wno-option-ignored -Wno-padded -Wno-return-std-move-in-c++11 -Wno-shorten-64-to-32 -Wno-sign-conversion -Wno-unknown-warning-option -Wno-unused-command-line-argument -Wno-weak-vtables -Wno-covered-switch-default -Wno-unused-result -Wno-unsafe-buffer-usage -Wno-deprecated-declarations -Wno-shadow-uncaptured-local -Wno-global-constructors -Wno-reserved-identifier -Wno-zero-as-null-pointer-constant -Wno-ignored-attributes -Wno-deprecated -Wno-incompatible-pointer-types -Wno-old-style-cast -Wno-unknown-attributes -Wno-microsoft-cpp-macro -Wno-microsoft-enum-value -Wno-language-extension-token -Wno-c++11-narrowing -Wno-float-equal -Wno-redundant-parens -Wno-format-nonliteral -Wno-unused-template -Wno-comma -Wno-suggest-destructor-override -Wno-switch-enum -Wno-shift-sign-overflow -Wno-suggest-override -Wno-inconsistent-missing-destructor-override -Wno-cast-function-type -Wno-nonportable-system-include-path -Wno-documentation -Wno-deprecated-builtins -Wno-enum-constexpr-conversion -Wno-unused-value -Wno-unused-parameter -Wno-missing-noreturn -Wno-tautological-constant-out-of-range-compare -x hip --offload-arch=gfx1100 -MD -MT src/CMakeFiles/MIOpen.dir/solver/conv_hip_implicit_gemm_3d_grouped_fwd_xdlops.cpp.o -MF src/CMakeFiles/MIOpen.dir/solver/conv_hip_implicit_gemm_3d_grouped_fwd_xdlops.cpp.o.d -o src/CMakeFiles/MIOpen.dir/solver/conv_hip_implicit_gemm_3d_grouped_fwd_xdlops.cpp.o -c /mnt/arhiv/rocm/release/MIOpen-rocm-6.2.0/src/solver/conv_hip_implicit_gemm_3d_grouped_fwd_xdlops.cpp
/mnt/arhiv/rocm/release/MIOpen-rocm-6.2.0/src/solver/conv_hip_implicit_gemm_3d_grouped_fwd_xdlops.cpp:36:10: fatal error: 'ck/library/tensor_operation_instance/gpu/grouped_convolution_forward_bilinear.hpp' file not found
   36 | #include <ck/library/tensor_operation_instance/gpu/grouped_convolution_forward_bilinear.hpp>
      |          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 error generated when compiling for gfx1100.
[288/554] Building CXX object src/CMakeFiles/MIOpen.dir/solver/conv_hip_implicit_gemm_3d_grouped_wrw_xdlops.cpp.o
FAILED: src/CMakeFiles/MIOpen.dir/solver/conv_hip_implicit_gemm_3d_grouped_wrw_xdlops.cpp.o 
/opt/rocm/llvm/bin/clang++ -DBOOST_ALL_NO_LIB=1 -DBOOST_ATOMIC_DYN_LINK -DBOOST_ATOMIC_NO_LIB -DBOOST_FILESYSTEM_DYN_LINK -DBOOST_FILESYSTEM_NO_LIB -DHIP_COMPILER_FLAGS=" -x hip   -D__HIP_PLATFORM_AMD__=1  -DUSE_PROF_API=1 --hip-link   -fno-offload-uniform-block " -DMIOPEN_BETA_API=1 -DMIOpen_EXPORTS -DUSE_PROF_API=1 -D__HIP_PLATFORM_AMD__=1 -I/mnt/arhiv/rocm/release/MIOpen-rocm-6.2.0/src/include -I/mnt/arhiv/rocm/rocm-build/build/miopen/include -I/mnt/arhiv/rocm/release/MIOpen-rocm-6.2.0/include -isystem /usr/include/half -isystem /opt/rocm/include/ck -isystem /opt/rocm/include/ck/utility -isystem /opt/rocm/include/ck/tensor_description -isystem /opt/rocm/include/ck/tensor -isystem /opt/rocm/include/ck/problem_transform -isystem /opt/rocm/include/ck/tensor_operation/gpu/device -isystem /opt/rocm/include/ck/tensor_operation/gpu/device/impl -isystem /opt/rocm/include/ck/tensor_operation/gpu/grid -isystem /opt/rocm/include/ck/tensor_operation/gpu/block -isystem /opt/rocm/include/ck/tensor_operation/gpu/warp -isystem /opt/rocm/include/ck/tensor_operation/gpu/thread -isystem /opt/rocm/include/ck/tensor_operation/gpu/element -isystem /opt/rocm/include/ck/library/utility -isystem /opt/rocm/include/ck/library/tensor_operation_instance -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/quantization -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/softmax -isystem /opt/rocm/include -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/conv_tensor_rearrange -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_data -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_weight -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/contraction -isystem /opt/rocm/include/ck/library/tensor_operation_instance/gpu/reduce -isystem /usr/include/eigen3 -isystem /opt/rocm/include/rocmlir -isystem /opt/rocm/include/rocmlir/external/include -fcf-protection=none -DNDEBUG -O3 -DNDEBUG -s -std=c++17 -fPIC -fvisibility-inlines-hidden   -U__HCC__ -Wall -Wextra -Wcomment -Wendif-labels -Wformat -Winit-self -Wreturn-type -Wsequence-point -Wswitch -Wtrigraphs -Wundef -Wuninitialized -Wunreachable-code -Wunused -Wno-ignored-qualifiers -Wno-sign-compare -Weverything -Wno-c++98-compat -Wno-c++98-compat-pedantic -Wno-conversion -Wno-double-promotion -Wno-exit-time-destructors -Wno-extra-semi -Wno-extra-semi-stmt -Wno-float-conversion -Wno-gnu-anonymous-struct -Wno-gnu-zero-variadic-macro-arguments -Wno-missing-prototypes -Wno-nested-anon-types -Wno-option-ignored -Wno-padded -Wno-return-std-move-in-c++11 -Wno-shorten-64-to-32 -Wno-sign-conversion -Wno-unknown-warning-option -Wno-unused-command-line-argument -Wno-weak-vtables -Wno-covered-switch-default -Wno-unused-result -Wno-unsafe-buffer-usage -Wno-deprecated-declarations -Wno-shadow-uncaptured-local -Wno-global-constructors -Wno-reserved-identifier -Wno-zero-as-null-pointer-constant -Wno-ignored-attributes -Wno-deprecated -Wno-incompatible-pointer-types -Wno-old-style-cast -Wno-unknown-attributes -Wno-microsoft-cpp-macro -Wno-microsoft-enum-value -Wno-language-extension-token -Wno-c++11-narrowing -Wno-float-equal -Wno-redundant-parens -Wno-format-nonliteral -Wno-unused-template -Wno-comma -Wno-suggest-destructor-override -Wno-switch-enum -Wno-shift-sign-overflow -Wno-suggest-override -Wno-inconsistent-missing-destructor-override -Wno-cast-function-type -Wno-nonportable-system-include-path -Wno-documentation -Wno-deprecated-builtins -Wno-enum-constexpr-conversion -Wno-unused-value -Wno-unused-parameter -Wno-missing-noreturn -Wno-tautological-constant-out-of-range-compare -x hip --offload-arch=gfx1100 -MD -MT src/CMakeFiles/MIOpen.dir/solver/conv_hip_implicit_gemm_3d_grouped_wrw_xdlops.cpp.o -MF src/CMakeFiles/MIOpen.dir/solver/conv_hip_implicit_gemm_3d_grouped_wrw_xdlops.cpp.o.d -o src/CMakeFiles/MIOpen.dir/solver/conv_hip_implicit_gemm_3d_grouped_wrw_xdlops.cpp.o -c /mnt/arhiv/rocm/release/MIOpen-rocm-6.2.0/src/solver/conv_hip_implicit_gemm_3d_grouped_wrw_xdlops.cpp
/mnt/arhiv/rocm/release/MIOpen-rocm-6.2.0/src/solver/conv_hip_implicit_gemm_3d_grouped_wrw_xdlops.cpp:36:10: fatal error: 'ck/library/tensor_operation_instance/gpu/grouped_convolution_backward_weight_bilinear.hpp' file not found
   36 | #include <ck/library/tensor_operation_instance/gpu/grouped_convolution_backward_weight_bilinear.hpp>
      |          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 error generated when compiling for gfx1100.
[304/554] Building CXX object src/CMakeFiles/MIOpen.dir/__/kernel.cpp.o
ninja: build stopped: subcommand failed.

The problem probably is comming from composable_kernel version 6.2.0 it's not compiled

[bghimireamd]

can you try building in latest MIOpen docker?