compiled version of cross gpu connection

.github/CODEOWNERS

-* @junliume @illsilin @carlushuang @aosewski @poyenc @geyyer @bartekxk
+* @junliume @illsilin @carlushuang @qianfengz @aosewski @poyenc @geyyer @bartekxk @andriy-ca
 # Documentation files
-docs/ @ROCm/rocm-documentation @junliume @illsilin @carlushuang @aosewski @poyenc @geyyer @bartekxk
-*.md @ROCm/rocm-documentation @junliume @illsilin @carlushuang @aosewski @poyenc @geyyer @bartekxk
-*.rst @ROCm/rocm-documentation @junliume @illsilin @carlushuang @aosewski @poyenc @geyyer @bartekxk
-.readthedocs.yaml @ROCm/rocm-documentation @junliume @illsilin @carlushuang @aosewski @poyenc @geyyer @bartekxk
+docs/ @ROCm/rocm-documentation @junliume @illsilin @carlushuang @qianfengz @aosewski @poyenc @geyyer @bartekxk @andriy-ca
+*.md @ROCm/rocm-documentation @junliume @illsilin @carlushuang @qianfengz @aosewski @poyenc @geyyer @bartekxk @andriy-ca
+*.rst @ROCm/rocm-documentation @junliume @illsilin @carlushuang @qianfengz @aosewski @poyenc @geyyer @bartekxk @andriy-ca
+.readthedocs.yaml @ROCm/rocm-documentation @junliume @illsilin @carlushuang @qianfengz @aosewski @poyenc @geyyer @bartekxk @andriy-ca
 # Header directory for Doxygen documentation
-library/include/ @ROCm/rocm-documentation @junliume @illsilin @carlushuang @aosewski @poyenc @geyyer @bartekxk
+library/include/ @ROCm/rocm-documentation @junliume @illsilin @carlushuang @qianfengz @aosewski @poyenc @geyyer @bartekxk @andriy-ca

.github/CONTRIBUTING.md

+We'd love for you to contribute to our source code!
+
+Some helpful links:
+
+- [Code of Conduct guidelines](https://www.contributor-covenant.org/version/2/1/code_of_conduct/code_of_conduct.txt)
+- [New issue guidelines](https://github.com/rocm/composable_kernel/blob/develop/.github/ISSUE_TEMPLATE.md)
+- [Submitting a pull request guidelines](https://github.com/rocm/composable_kernel/blob/develop/.github/PULL_REQUEST_TEMPLATE.md)
+- [Maintainers](https://github.com/rocm/composable_kernel/blob/develop/CONTRIBUTORS.md)
+- [General information](https://github.com/rocm/composable_kernel/blob/develop/README.md)
+- [ROCm documentation](https://rocm.docs.amd.com/en/latest/how-to/llm-fine-tuning-optimization/optimizing-with-composable-kernel.html)
\ No newline at end of file

.github/ISSUE_TEMPLATE.md

+When creating an issue, please check if a similar issue already exists.
+
+### When reporting a bug, please include:
+- [ ] A descriptive title
+- [ ] An isolated way to reproduce the behavior (preferably a docker container with a repro)
+- [ ] ROCm version, clang version, Composable Kernel commit pin
+- [ ] Environment variables
+- [ ] The behavior you expect to see, and the behavior you actually see
+
+### When requesting a feature, please include:
+- [ ] A descriptive title
+- [ ] A detailed description of the problem you are trying to solve
+- [ ] An overview of the suggested solution
+- [ ] Explanation why the solution is an improvement
\ No newline at end of file

.github/PULL_REQUEST_TEMPLATE.md

+## Proposed changes
+
+Please describe the motivation behind the pull request, whether it enables a new feature or fixes a bug. If there are associated pull requests or issues, please link them to the pull request.
+
+## Checklist
+
+Please put an `x` into the boxes that apply. You can also fill these out after creating the PR. If you're not sure, please don't hesitate to ask.
+
+- [ ] I have added tests relevant to the introduced functionality, and the unit tests are passing locally
+- [ ] I have added inline documentation which enables the maintainers with understanding the motivation
+- [ ] I have removed the stale documentation which is no longer relevant after this pull request
+- [ ] (If this change is user-facing) I have added release notes which provide the end users with a brief summary of the improvement from this pull request
+- [ ] I have run `clang-format` on all changed files
+- [ ] Any dependent changes have been merged
+
+## Discussion
+
+If this is a relatively large or complex change, feel free to start a discussion by explaining why you chose the solution you did and what alternatives you considered
+

CMakeLists.txt

@@ -185,14 +185,29 @@ if (SUPPORTED_GPU_TARGETS MATCHES "gfx9")
     add_definitions(-DCK_USE_XDL)
     set(CK_USE_XDL "ON")
 endif()
+if (SUPPORTED_GPU_TARGETS MATCHES "gfx94")
+    message("Enabling FP8 gemms on native architectures")
+    add_definitions(-DCK_USE_GFX94)
+    set(CK_USE_GFX94 "ON")
+endif()
 if (SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
     message("Enabling WMMA instances")
     add_definitions(-DCK_USE_WMMA)
     set(CK_USE_WMMA "ON")
 endif()
+if (SUPPORTED_GPU_TARGETS MATCHES "gfx12")
+    add_definitions(-DCK_USE_OCP_FP8)
+    set(CK_USE_OCP_FP8 "ON")
+endif()
+if (SUPPORTED_GPU_TARGETS MATCHES "gfx90a" OR SUPPORTED_GPU_TARGETS MATCHES "gfx94")
+    add_definitions(-DCK_USE_FNUZ_FP8)
+    set(CK_USE_FNUZ_FP8 "ON")
+endif()
+
 option(CK_USE_FP8_ON_UNSUPPORTED_ARCH "Enable FP8 GEMM instances on older architectures" OFF)
 if(CK_USE_FP8_ON_UNSUPPORTED_ARCH AND (SUPPORTED_GPU_TARGETS MATCHES "gfx90a" OR SUPPORTED_GPU_TARGETS MATCHES "gfx908"))
     add_definitions(-DCK_USE_FP8_ON_UNSUPPORTED_ARCH)
+    set(CK_USE_FP8_ON_UNSUPPORTED_ARCH "ON")
 endif()
 
 # CK config file to record supported datatypes, etc.
@@ -221,7 +236,7 @@ if(NOT WIN32 AND ${hip_VERSION_FLAT} GREATER 600140090)
 endif()
 set(check-coerce)
 check_cxx_compiler_flag(" -mllvm -amdgpu-coerce-illegal-types=1" check-coerce)
-if(NOT WIN32 AND check-coerce AND ${hip_VERSION_FLAT} GREATER 600241132 AND ${hip_VERSION_FLAT} LESS 600300000)
+if(NOT WIN32 AND check-coerce AND ${hip_VERSION_FLAT} GREATER 600241132)
    message("Adding the amdgpu-coerce-illegal-types=1")
    add_compile_options("SHELL: -mllvm -amdgpu-coerce-illegal-types=1")
 endif()

CONTRIBUTORS.md

+[Back to the main page](./README.md)
 # Composable Kernel Developers and Contributors
 
 This is the list of developers and contributors to Composable Kernel library

Dockerfile

-FROM ubuntu:20.04
+FROM ubuntu:22.04
 ARG DEBIAN_FRONTEND=noninteractive
-ARG ROCMVERSION=6.2
+ARG ROCMVERSION=6.3
 ARG compiler_version=""
 ARG compiler_commit=""
 ARG CK_SCCACHE=""
-
-RUN set -xe
-
 ARG DEB_ROCM_REPO=http://repo.radeon.com/rocm/apt/.apt_$ROCMVERSION/
-RUN useradd -rm -d /home/jenkins -s /bin/bash -u 1004 jenkins
-# Add rocm repository
-RUN chmod 1777 /tmp
-RUN apt-get update
-RUN apt-get install -y --allow-unauthenticated apt-utils wget gnupg2 curl
-
 ENV APT_KEY_DONT_WARN_ON_DANGEROUS_USAGE=DontWarn
-RUN curl -fsSL https://repo.radeon.com/rocm/rocm.gpg.key | gpg --dearmor -o /etc/apt/trusted.gpg.d/rocm-keyring.gpg
 
-RUN if [ "$ROCMVERSION" != "6.3" ]; then \
-        sh -c "wget https://repo.radeon.com/amdgpu-install/$ROCMVERSION/ubuntu/focal/amdgpu-install_6.2.60200-1_all.deb  --no-check-certificate" && \
-        apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-unauthenticated ./amdgpu-install_6.2.60200-1_all.deb && \
+# Add rocm repository
+RUN set -xe && \
+    useradd -rm -d /home/jenkins -s /bin/bash -u 1004 jenkins && \
+    apt-get update && apt-get install -y --allow-unauthenticated apt-utils wget gnupg2 curl && \
+    curl -fsSL https://repo.radeon.com/rocm/rocm.gpg.key | gpg --dearmor -o /etc/apt/trusted.gpg.d/rocm-keyring.gpg
+
+RUN if [ "$ROCMVERSION" != "6.4" ]; then \
+        sh -c "wget https://repo.radeon.com/amdgpu-install/$ROCMVERSION/ubuntu/focal/amdgpu-install_6.3.60300-1_all.deb  --no-check-certificate" && \
+        apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-unauthenticated ./amdgpu-install_6.3.60300-1_all.deb && \
         wget -qO - http://repo.radeon.com/rocm/rocm.gpg.key | apt-key add - && \
         sh -c "echo deb [arch=amd64 signed-by=/etc/apt/trusted.gpg.d/rocm-keyring.gpg] $DEB_ROCM_REPO focal main > /etc/apt/sources.list.d/rocm.list" && \
         sh -c 'echo deb [arch=amd64 signed-by=/etc/apt/trusted.gpg.d/rocm-keyring.gpg] https://repo.radeon.com/amdgpu/$ROCMVERSION/ubuntu focal main > /etc/apt/sources.list.d/amdgpu.list'; \
-    elif [ "$ROCMVERSION" = "6.3" ] && [ "$compiler_version" = "rc1" ]; then \
-        sh -c "wget http://artifactory-cdn.amd.com/artifactory/list/amdgpu-deb/amdgpu-install-internal_6.3.0.1-20.04-1_all.deb --no-check-certificate" && \
-        apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install dialog libpopt0 rsync && DEBIAN_FRONTEND=noninteractive apt-get install ./amdgpu-install-internal_6.3.0.1-20.04-1_all.deb && \
-        sh -c 'echo deb [arch=amd64 trusted=yes] http://compute-artifactory.amd.com/artifactory/list/rocm-release-archive-20.04-deb/ 6.3.0.1 rel-5 > /etc/apt/sources.list.d/rocm-build.list' && \
-        amdgpu-repo --amdgpu-build=2033700; \
     fi
 
-RUN sh -c "echo deb http://mirrors.kernel.org/ubuntu focal main universe | tee -a /etc/apt/sources.list"
-RUN amdgpu-install -y --usecase=rocm --no-dkms
+RUN sh -c "echo deb http://mirrors.kernel.org/ubuntu focal main universe | tee -a /etc/apt/sources.list" && \
+    amdgpu-install -y --usecase=rocm --no-dkms
 
 ## Sccache binary built from source for ROCm, only install if CK_SCCACHE is defined
 ARG SCCACHE_REPO_URL=http://compute-artifactory.amd.com/artifactory/rocm-generic-experimental/rocm-sccache
@@ -57,6 +48,7 @@ RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-
     libnuma-dev \
     libpthread-stubs0-dev \
     llvm-amdgpu \
+    mpich \
     net-tools \
     pkg-config \
     python \
@@ -72,72 +64,52 @@ RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-
     nano \
     zlib1g-dev \
     zip \
+    libzstd-dev \
     openssh-server \
     clang-format-12 \
     kmod && \
     apt-get clean && \
-    rm -rf /var/lib/apt/lists/*
+    rm -rf /var/lib/apt/lists/* && \
+    rm -rf amdgpu-install* && \
+# Remove unnecessary rocm components that take a lot of space
+    apt-get remove -y rocblas rocfft rocsparse composablekernel-dev hipblaslt
 
-# hipTensor requires rocm-llvm-dev for rocm versions > 6.0.1
-RUN if [ "$ROCMVERSION" = "6.1" ]; then \
-        sh -c "apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-unauthenticated rocm-llvm-dev"; \
-    fi
 # Update the cmake to version 3.27.5
-RUN pip install --upgrade cmake==3.27.5
-
+RUN pip install --upgrade cmake==3.27.5 && \
 #Install latest ccache
-RUN git clone https://github.com/ccache/ccache.git && \
-    cd ccache && mkdir build && cd build && cmake .. && make install
-
+    git clone https://github.com/ccache/ccache.git && \
+    cd ccache && mkdir build && cd build && cmake .. && make install && \
 #Install ninja build tracing tools
-RUN wget -qO /usr/local/bin/ninja.gz https://github.com/ninja-build/ninja/releases/latest/download/ninja-linux.zip
-RUN gunzip /usr/local/bin/ninja.gz
-RUN chmod a+x /usr/local/bin/ninja
-RUN git clone https://github.com/nico/ninjatracing.git
-
+    cd / && \
+    wget -qO /usr/local/bin/ninja.gz https://github.com/ninja-build/ninja/releases/latest/download/ninja-linux.zip && \
+    gunzip /usr/local/bin/ninja.gz && \
+    chmod a+x /usr/local/bin/ninja && \
+    git clone https://github.com/nico/ninjatracing.git && \
 #Install latest cppcheck
-RUN git clone https://github.com/danmar/cppcheck.git && \
-    cd cppcheck && mkdir build && cd build && cmake .. && cmake --build .
-WORKDIR /
-
-# Setup ubsan environment to printstacktrace
-RUN ln -s /usr/bin/llvm-symbolizer-3.8 /usr/local/bin/llvm-symbolizer
-ENV UBSAN_OPTIONS=print_stacktrace=1
-
+    git clone https://github.com/danmar/cppcheck.git && \
+    cd cppcheck && mkdir build && cd build && cmake .. && cmake --build . && \
+    cd / && \
 # Install an init system
-RUN wget https://github.com/Yelp/dumb-init/releases/download/v1.2.0/dumb-init_1.2.0_amd64.deb
-RUN dpkg -i dumb-init_*.deb && rm dumb-init_*.deb
-
-ARG PREFIX=/opt/rocm
+    wget https://github.com/Yelp/dumb-init/releases/download/v1.2.0/dumb-init_1.2.0_amd64.deb && \
+    dpkg -i dumb-init_*.deb && rm dumb-init_*.deb && \
 # Install packages for processing the performance results
-RUN pip3 install --upgrade pip
-RUN pip3 install sqlalchemy==1.4.46
-RUN pip3 install pymysql
-RUN pip3 install pandas==2.0.3
-RUN pip3 install setuptools-rust
-RUN pip3 install sshtunnel==0.4.0
-# Setup ubsan environment to printstacktrace
-ENV UBSAN_OPTIONS=print_stacktrace=1
-
-ENV LC_ALL=C.UTF-8
-ENV LANG=C.UTF-8
-RUN groupadd -f render
-
+    pip3 install --upgrade pip && \
+    pip3 install sqlalchemy==2.0.36 pymysql pandas==2.2.3 setuptools-rust sshtunnel==0.4.0 && \
+# Add render group
+    groupadd -f render && \
 # Install the new rocm-cmake version
-RUN git clone -b master https://github.com/ROCm/rocm-cmake.git  && \
+    git clone -b master https://github.com/ROCm/rocm-cmake.git  && \
     cd rocm-cmake && mkdir build && cd build && \
     cmake  .. && cmake --build . && cmake --build . --target install
 
 WORKDIR /
-
+# Add alternative compilers, if necessary
 ENV compiler_version=$compiler_version
 ENV compiler_commit=$compiler_commit
-RUN sh -c "echo compiler version = '$compiler_version'"
-RUN sh -c "echo compiler commit = '$compiler_commit'"
-
-ARG DISABLE_CACHE=0
+RUN sh -c "echo compiler version = '$compiler_version'" && \
+    sh -c "echo compiler commit = '$compiler_commit'"
 
-RUN if ( [ "$compiler_version" = "amd-staging" ] || [ "$compiler_version" = "amd-mainline-open" ] ) && [ "$compiler_commit" = "" ]; then \
+RUN if ( [ "$compiler_version" = "amd-staging" ] || [ "$compiler_version" = "amd-mainline" ] ) && [ "$compiler_commit" = "" ]; then \
         git clone -b "$compiler_version" https://github.com/ROCm/llvm-project.git && \
         cd llvm-project && mkdir build && cd build && \
         cmake -DCMAKE_INSTALL_PREFIX=/opt/rocm/llvm -DCMAKE_BUILD_TYPE=Release -DLLVM_ENABLE_ASSERTIONS=1 -DLLVM_TARGETS_TO_BUILD="AMDGPU;X86" -DLLVM_ENABLE_PROJECTS="clang;lld" -DLLVM_ENABLE_RUNTIMES="compiler-rt" ../llvm && \
@@ -145,16 +117,10 @@ RUN if ( [ "$compiler_version" = "amd-staging" ] || [ "$compiler_version" = "amd
     else echo "using the release compiler"; \
     fi
 
-RUN if ( [ "$compiler_version" = "amd-staging" ] || [ "$compiler_version" = "amd-mainline-open" ] ) && [ "$compiler_commit" != "" ]; then \
+RUN if ( [ "$compiler_version" = "amd-staging" ] || [ "$compiler_version" = "amd-mainline" ] ) && [ "$compiler_commit" != "" ]; then \
         git clone -b "$compiler_version" https://github.com/ROCm/llvm-project.git && \
         cd llvm-project && git checkout "$compiler_commit" && echo "checking out commit $compiler_commit" && mkdir build && cd build && \
         cmake -DCMAKE_INSTALL_PREFIX=/opt/rocm/llvm -DCMAKE_BUILD_TYPE=Release -DLLVM_ENABLE_ASSERTIONS=1 -DLLVM_TARGETS_TO_BUILD="AMDGPU;X86" -DLLVM_ENABLE_PROJECTS="clang;lld" -DLLVM_ENABLE_RUNTIMES="compiler-rt" ../llvm && \
         make -j 8 ; \
     else echo "using the release compiler"; \
     fi
-
-#clean-up the deb package
-RUN sh -c "rm -rf amdgpu-install*"
-
-#ENV HIP_CLANG_PATH='/llvm-project/build/bin'
-#RUN sh -c "echo HIP_CLANG_PATH = '$HIP_CLANG_PATH'"

Dockerfile.compiler

+ARG BASE_DOCKER="rocm/composable_kernel:ck_ub22.04_rocm6.3"
+FROM $BASE_DOCKER
+ARG compiler_version=""
+ARG compiler_commit=""
+
+# Add alternative compilers, if necessary
+ENV compiler_version=$compiler_version
+ENV compiler_commit=$compiler_commit
+RUN sh -c "echo compiler version = '$compiler_version'" && \
+    sh -c "echo compiler commit = '$compiler_commit'"
+
+RUN if ( [ "$compiler_version" = "amd-staging" ] || [ "$compiler_version" = "amd-mainline" ] ) && [ "$compiler_commit" = "" ]; then \
+        git clone -b "$compiler_version" https://github.com/ROCm/llvm-project.git && \
+        cd llvm-project && mkdir build && cd build && \
+        cmake -DCMAKE_INSTALL_PREFIX=/opt/rocm/llvm -DCMAKE_BUILD_TYPE=Release -DLLVM_ENABLE_ASSERTIONS=1 -DLLVM_TARGETS_TO_BUILD="AMDGPU;X86" -DLLVM_ENABLE_PROJECTS="clang;lld" -DLLVM_ENABLE_RUNTIMES="compiler-rt" ../llvm && \
+        make -j 16 ; \
+    else echo "using the release compiler"; \
+    fi
+
+RUN if ( [ "$compiler_version" = "amd-staging" ] || [ "$compiler_version" = "amd-mainline" ] ) && [ "$compiler_commit" != "" ]; then \
+        git clone -b "$compiler_version" https://github.com/ROCm/llvm-project.git && \
+        cd llvm-project && git checkout "$compiler_commit" && echo "checking out commit $compiler_commit" && mkdir build && cd build && \
+        cmake -DCMAKE_INSTALL_PREFIX=/opt/rocm/llvm -DCMAKE_BUILD_TYPE=Release -DLLVM_ENABLE_ASSERTIONS=1 -DLLVM_TARGETS_TO_BUILD="AMDGPU;X86" -DLLVM_ENABLE_PROJECTS="clang;lld" -DLLVM_ENABLE_RUNTIMES="compiler-rt" ../llvm && \
+        make -j 16 ; \
+    else echo "using the release compiler"; \
+    fi

README.md

@@ -26,23 +26,15 @@ The current CK library is structured into four layers:
 
 ## General information
 
-To build our documentation locally, use the following code:
-
-``` bash
-cd docs
-pip3 install -r sphinx/requirements.txt
-python3 -m sphinx -T -E -b html -d _build/doctrees -D language=en . _build/html
-```
-
-You can find a list of our developers and contributors on our [Contributors](/CONTRIBUTORS.md) page.
-
-```note
-If you use CK, cite us as follows:
-
-* [Realizing Tensor Operators Using Coordinate Transformations and Tile Based Programming](???):
-  This paper will be available on arXiv soon.
-* [CITATION.cff](/CITATION.cff)
-```
+* [CK supported operations](include/ck/README.md)
+* [CK Tile supported operations](include/ck_tile/README.md)
+* [CK wrapper](client_example/25_wrapper/README.md)
+* [CK codegen](codegen/README.md)
+* [CK profiler](profiler/README.md)
+* [Examples (Custom use of CK supported operations)](example/README.md)
+* [Client examples (Use of CK supported operations with instance factory)](client_example/README.md)
+* [Terminology](/TERMINOLOGY.md)
+* [Contributors](/CONTRIBUTORS.md)
 
 CK is released under the **[MIT license](/LICENSE)**.
 
@@ -137,6 +129,14 @@ Docker images are available on [DockerHub](https://hub.docker.com/r/rocm/composa
 
     You can find instructions for running ckProfiler in [profiler](/profiler).
 
+* Build our documentation locally:
+
+    ``` bash
+    cd docs
+    pip3 install -r sphinx/requirements.txt
+    python3 -m sphinx -T -E -b html -d _build/doctrees -D language=en . _build/html
+    ```
+
 Note the `-j` option for building with multiple threads in parallel, which speeds up the build significantly.
 However, `-j` launches unlimited number of threads, which can cause the build to run out of memory and
 crash. On average, you should expect each thread to use ~2Gb of RAM.
@@ -154,8 +154,7 @@ Additional cmake flags can be used to significantly speed-up the build:
   other platforms have faster instances, such as `xdl` or `wmma`, available.
 
 * `CK_USE_FP8_ON_UNSUPPORTED_ARCH` (default is OFF) must be set to ON in order to build instances,
-  such as `gemm_universal` and `gemm_multiply_multiply` for fp8 data type for GPU targets which do not
-  have native support for fp8 data type, such as gfx908 or gfx90a. These instances are useful on
+  such as `gemm_universal`, `gemm_universal_streamk` and `gemm_multiply_multiply` for fp8 data type for GPU targets which do not  have native support for fp8 data type, such as gfx908 or gfx90a. These instances are useful on
   architectures like the MI100/MI200 for the functional support only.
 
 ## Using sccache for building

TERMINOLOGY.md

+[Back to the main page](./README.md)
+# Composable Kernel terminology
\ No newline at end of file

client_example/24_grouped_conv_activation/CMakeLists.txt

@@ -54,7 +54,7 @@ target_link_libraries(client_conv3d_fwd_convscale_relu_amax_fp8
                       PRIVATE composable_kernel::device_conv_operations
                               composable_kernel::device_other_operations
                               composable_kernel::device_reduction_operations
-                              utility)
+                              composable_kernel::utility)
 # Fwd convscale + AMAX
 add_executable(client_conv3d_fwd_convscale_amax_fp8
                grouped_convnd_fwd_convscale_reduce/conv3d_fwd_convscale_amax_fp8.cpp)
@@ -62,7 +62,7 @@ target_link_libraries(client_conv3d_fwd_convscale_amax_fp8
                       PRIVATE composable_kernel::device_conv_operations
                               composable_kernel::device_other_operations
                               composable_kernel::device_reduction_operations
-                              utility)
+                              composable_kernel::utility)
 # Fwd convscale
 add_executable(client_conv3d_fwd_convscale_fp8
                grouped_convnd_fwd_convscale/conv3d_fwd_convscale_fp8.cpp)

client_example/25_wrapper/README.md

+[Back to the main page](../../README.md)
 # Composable Kernel wrapper GEMM tutorial
 
-This tutorial demonstrates how to implement matrix multiplication using Composable Kernel (CK)
-wrapper. We present the base version of GEMM without most of the available optimizations; however,
-it's worth noting that CK has kernels with different optimizations.
+This tutorial demonstrates how to implement matrix multiplication using Composable Kernel (CK) wrapper. We present the base version of GEMM without most of the available optimizations; however, it's worth noting that CK has kernels with different optimizations.
 
-To implement these optimizations, you can use the CK wrapper or directly use available instances in
-CK. You can also refer to the
-[optimized GEMM example](https://github.com/ROCm/composable_kernel/blob/develop/client_example/25_wrapper/wrapper_optimized_gemm.cpp),
-that uses CK wrapper based on the
-[`gridwise_gemm_xdlops_v2r3`](https://github.com/ROCm/composable_kernel/blob/develop/include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp) implementation.
+To implement these optimizations, you can use the CK wrapper or directly use available instances in CK. You can also refer to the [optimized GEMM example](https://github.com/ROCm/composable_kernel/blob/develop/client_example/25_wrapper/wrapper_optimized_gemm.cpp), that uses CK wrapper based on the [`gridwise_gemm_xdlops_v2r3`](https://github.com/ROCm/composable_kernel/blob/develop/include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp) implementation.
 
 The kernel definition should look similar to:
 

client_example/31_grouped_gemm_bf16Aint8B/grouped_gemm_multiply_bias_fastgelu_xdl_bf16_i8.cpp

@@ -121,7 +121,7 @@ bool run_grouped_gemm(const ProblemSize& problem_size, const ExecutionConfig& co
     constexpr ck::index_t NumDTensor = 2;
 
     using GroupedGemmKernelArgument =
-        ck::tensor_operation::device::GroupedGemmTileLoopKernelArguments<NumDTensor>;
+        ck::tensor_operation::device::GroupedGemmKernelArgument<NumDTensor>;
 
     std::vector<GroupedGemmKernelArgument> grouped_gemm_kernel_args_;
     grouped_gemm_kernel_args_.reserve(group_count);

client_example/31_grouped_gemm_bf16Aint8B/grouped_gemm_multiply_xdl_bf16_i8.cpp

@@ -120,7 +120,7 @@ bool run_grouped_gemm(const ProblemSize& problem_size, const ExecutionConfig& co
     constexpr ck::index_t NumDTensor = 1;
 
     using GroupedGemmKernelArgument =
-        ck::tensor_operation::device::GroupedGemmTileLoopKernelArguments<NumDTensor>;
+        ck::tensor_operation::device::GroupedGemmKernelArgument<NumDTensor>;
 
     std::vector<GroupedGemmKernelArgument> grouped_gemm_kernel_args_;
     grouped_gemm_kernel_args_.reserve(group_count);

client_example/CMakeLists.txt

@@ -56,13 +56,21 @@ if (GPU_TARGETS)
         add_definitions(-DCK_USE_WMMA)
         set(CK_USE_WMMA "ON")
     endif()
+    if (GPU_TARGETS MATCHES "gfx12")
+        add_definitions(-DCK_USE_OCP_FP8)
+        set(CK_USE_OCP_FP8 "ON")
+    endif()
+    if (GPU_TARGETS MATCHES "gfx90a" OR GPU_TARGETS MATCHES "gfx94")
+        add_definitions(-DCK_USE_FNUZ_FP8)
+        set(CK_USE_FNUZ_FP8 "ON")
+    endif()
 else()
     add_definitions(-DCK_USE_WMMA -DCK_USE_XDL)
     set(CK_USE_XDL "ON")
     set(CK_USE_WMMA "ON")
 endif()
 
-find_package(composable_kernel COMPONENTS device_other_operations device_gemm_operations device_conv_operations  device_reduction_operations)
+find_package(composable_kernel COMPONENTS device_other_operations device_gemm_operations device_conv_operations  device_reduction_operations utility)
 if(GPU_TARGETS MATCHES "gfx9")
     find_package(composable_kernel COMPONENTS device_contraction_operations)
 endif()

client_example/README.md

+[Back to the main page](../README.md)
+# Composable Kernel client examples
 ##
 Client application links to CK library, and therefore CK library needs to be installed before building client applications.
 

codegen/CMakeLists.txt

@@ -7,6 +7,7 @@ set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/lib)
 set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/lib)
 set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/bin)
 set(CK_ROOT ${CMAKE_CURRENT_SOURCE_DIR}/..)
+configure_file(${CK_ROOT}/include/ck/config.h.in ${CK_ROOT}/include/ck/config.h)
 
 find_package(ROCM)
 include(ROCMInstallTargets)

codegen/README.md

+[Back to the main page](../README.md)
+# Composable Kernel codegen
\ No newline at end of file

codegen/test/rtc/include/rtc/hip.hpp

@@ -4,6 +4,7 @@
 #include <hip/hip_runtime_api.h>
 #include <memory>
 #include <string>
+#include <stdexcept>
 
 namespace rtc {