merge from develop and revisison for pr#881

.pre-commit-config.yaml

+repos:
+-   repo: local
+    hooks:
+    -   id: clang-format
+        name: clang-format
+        entry: clang-format-12 -i --style=file
+        language: system
+        types_or: [c++, inc]
+    -   id: copyright-year-checker
+        name: copyright-year-checker
+        entry: script/check_copyright_year.sh
+        verbose: false
+        language: script
+        types: [c++]

CHANGELOG.md

@@ -12,6 +12,11 @@ Full documentation for Composable Kernel is not yet available.
 - Improve proformance of normalization kernel
 
 ### Added
+- Added new cmake flag "DL_KERNELS" must be set to "ON" in order to build the gemm_dl and batched_gemm_multi_d_dl instances.
+- Added new cmake flag "DTYPES" which could be set to any subset of "fp64;fp32;fp16;fp8;bf16;int8" to build instance of select data types.
+- Added new cmake flag "INSTANCES_ONLY" which will only build CK library and instances without the tests, examples, or profiler.
+- Added new feature: if GPU_TARGETS is not set on cmake command line, CK will be built for all targets supported by compiler.
+- Added support on MI300A/MI300X.
 - Added support on NAVI3x.
 - Added user tutorial (#563).
 - Added more instances for irregular GEMM sizes (#560).
@@ -20,6 +25,8 @@ Full documentation for Composable Kernel is not yet available.
 - Added multi-embeddings support (#542).
 - Added Navi3x blockwise GEMM and real GEMM support (#541).
 - Added Navi grouped ConvBwdWeight support (#505).
+- Added MaxPool, AvgPool forward (#815).
+- Added MaxPool backward (#750).
 
 ### Changed
 - Changed ...

CMakeLists.txt

 cmake_minimum_required(VERSION 3.14)
 
+set(version 1.1.0)
 # Check support for CUDA/HIP in Cmake
-project(composable_kernel)
+project(composable_kernel VERSION ${version})
 
 list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake")
 
+if (DTYPES)
+    add_definitions(-DDTYPES)
+    if (DTYPES MATCHES "int8")
+        add_definitions(-DCK_ENABLE_INT8)
+        set(CK_ENABLE_INT8 "ON")
+    endif()
+    if (DTYPES MATCHES "fp8")
+        add_definitions(-DCK_ENABLE_FP8)
+        set(CK_ENABLE_FP8 "ON")
+    endif()
+    if (DTYPES MATCHES "fp16")
+        add_definitions(-DCK_ENABLE_FP16)
+        set(CK_ENABLE_FP16 "ON")
+    endif()
+    if (DTYPES MATCHES "fp32")
+        add_definitions(-DCK_ENABLE_FP32)
+        set(CK_ENABLE_FP32 "ON")
+    endif()
+    if (DTYPES MATCHES "fp64")
+        add_definitions(-DCK_ENABLE_FP64)
+        set(CK_ENABLE_FP64 "ON")
+    endif()
+    if (DTYPES MATCHES "bf16")
+        add_definitions(-DCK_ENABLE_BF16)
+        set(CK_ENABLE_BF16 "ON")
+    endif()
+    message("DTYPES macro set to ${DTYPES}")
+else()
+    add_definitions(-DCK_ENABLE_INT8 -DCK_ENABLE_FP8 -DCK_ENABLE_FP16 -DCK_ENABLE_FP32 -DCK_ENABLE_FP64 -DCK_ENABLE_BF16)
+    set(CK_ENABLE_ALL_DTYPES "ON")
+endif()
+
+if(DL_KERNELS)
+    add_definitions(-DDL_KERNELS)
+    set(CK_ENABLE_DL_KERNELS "ON")
+endif()
+
+if(INSTANCES_ONLY)
+    add_definitions(-DINSTANCES_ONLY)
+    set(CK_ENABLE_INSTANCES_ONLY "ON")
+endif()
+
+# CK config file to record supported datatypes, etc.
+configure_file("${PROJECT_SOURCE_DIR}/include/ck/config.h.in" "${PROJECT_BINARY_DIR}/include/ck/config.h")
+
+# CK version file to record release version as well as git commit hash
+find_package(Git REQUIRED)
+execute_process(COMMAND "${GIT_EXECUTABLE}" rev-parse HEAD OUTPUT_VARIABLE COMMIT_ID OUTPUT_STRIP_TRAILING_WHITESPACE)
+configure_file("${PROJECT_SOURCE_DIR}/include/ck/version.h.in" "${PROJECT_BINARY_DIR}/include/ck/version.h")
+
 enable_testing()
 
 set(ROCM_SYMLINK_LIBS OFF)
@@ -16,11 +67,57 @@ include(ROCMSetupVersion)
 include(ROCMInstallSymlinks)
 include(ROCMCreatePackage)
 include(CheckCXXCompilerFlag)
-
-rocm_setup_version(VERSION 0.2.0)
+include(ROCMCheckTargetIds)
 include(TargetFlags)
+
+rocm_setup_version(VERSION ${version})
+
 list(APPEND CMAKE_PREFIX_PATH ${CMAKE_INSTALL_PREFIX} ${CMAKE_INSTALL_PREFIX}/llvm ${CMAKE_INSTALL_PREFIX}/hip /opt/rocm /opt/rocm/llvm /opt/rocm/hip)
 
+message("GPU_TARGETS= ${GPU_TARGETS}")
+
+message("checking which targets are supported")
+#This is the list of targets to be used in case GPU_TARGETS is not set on command line
+#These targets will be filtered and only supported ones will be used
+#Setting GPU_TARGETS on command line will override this list
+if(NOT PROFILER_ONLY)
+    rocm_check_target_ids(DEFAULT_GPU_TARGETS
+        TARGETS "gfx900;gfx906;gfx908;gfx90a;gfx940;gfx941;gfx942;gfx1030;gfx1100;gfx1101;gfx1102")
+else()
+    add_definitions(-DPROFILER_ONLY)
+    if(GPU_TARGETS)
+        message(FATAL_ERROR "For PROFILE_ONLY build, please do not set GPU_TARGETS, use GPU_ARCH = gfx9, gfx10, or gfx11")
+    endif()
+    if(GPU_ARCH MATCHES "gfx9")
+        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx900;gfx906;gfx908;gfx90a;gfx940;gfx941;gfx942")
+    elseif(GPU_ARCH MATCHES "gfx10")
+        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx1030")
+    elseif(GPU_ARCH MATCHES "gfx11")
+        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx1100;gfx1101;gfx1102")
+    else()
+        message(FATAL_ERROR "For PROFILE_ONLY build, please specify GPU_ARCH as gfx9, gfx10, or gfx11")
+    endif()
+endif()
+
+message("Supported GPU_TARGETS= ${DEFAULT_GPU_TARGETS}")
+
+set(AMDGPU_TARGETS "${DEFAULT_GPU_TARGETS}" CACHE STRING " ")
+
+if(GPU_TARGETS)
+    message("Building CK for the following targets: ${GPU_TARGETS}")
+else()
+    message("Building CK for the following targets: ${AMDGPU_TARGETS}")
+endif()
+find_package(hip)
+# No assumption that HIP kernels are launched with uniform block size for backward compatibility
+# SWDEV-413293 and https://reviews.llvm.org/D155213
+math(EXPR hip_VERSION_FLAT "(${hip_VERSION_MAJOR} * 1000 + ${hip_VERSION_MINOR}) * 100000 + ${hip_VERSION_PATCH}")
+message("hip_version_flat=${hip_VERSION_FLAT}")
+if(${hip_VERSION_FLAT} GREATER 500723302)
+   message("Adding the fno-offload-uniform-block compiler flag")
+   add_compile_options(-fno-offload-uniform-block)
+endif()
+
 option(USE_BITINT_EXTENSION_INT4, "Whether to enable clang's BitInt extension to provide int4 data type." OFF)
 option(USE_OPT_NAVI3X, "Whether to enable LDS cumode and Wavefront32 mode for NAVI3X silicons." OFF)
 
@@ -238,13 +335,14 @@ set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/lib)
 set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/lib)
 set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/bin)
 
+# set CK project include directories
 include_directories(BEFORE
+    ${PROJECT_BINARY_DIR}/include
     ${PROJECT_SOURCE_DIR}/include
     ${PROJECT_SOURCE_DIR}/library/include
     ${HIP_INCLUDE_DIRS}
 )
 
-
 SET(BUILD_DEV ON CACHE BOOL "BUILD_DEV")
 if(BUILD_DEV)
     add_compile_options(-Werror)
@@ -258,36 +356,80 @@ file(GLOB_RECURSE INSTANCE_FILES "${PROJECT_SOURCE_DIR}/*/device_*_instance.cpp"
 file(GLOB dir_list RELATIVE ${PROJECT_SOURCE_DIR}/library/src/tensor_operation_instance/gpu ${PROJECT_SOURCE_DIR}/library/src/tensor_operation_instance/gpu/*)
 set(CK_DEVICE_INSTANCES)
 FOREACH(subdir_path ${dir_list})
-    IF(IS_DIRECTORY "${PROJECT_SOURCE_DIR}/library/src/tensor_operation_instance/gpu/${subdir_path}")
-       list(APPEND CK_DEVICE_INSTANCES device_${subdir_path}_instance)
-    ENDIF()
+set(target_dir)
+IF(IS_DIRECTORY "${PROJECT_SOURCE_DIR}/library/src/tensor_operation_instance/gpu/${subdir_path}")
+    set(cmake_instance)
+    file(READ "${PROJECT_SOURCE_DIR}/library/src/tensor_operation_instance/gpu/${subdir_path}/CMakeLists.txt" cmake_instance)
+    set(add_inst 0)
+    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"fp8\" " AND DTYPES MATCHES "fp8")
+        #message("fp8 instance found!")
+        set(add_inst 1)
+    endif()
+    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"fp16\"" AND DTYPES MATCHES "fp16")
+        #message("fp16 instance found!")
+        set(add_inst 1)
+    endif()
+    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"fp32\"" AND DTYPES MATCHES "fp32")
+        #message("fp32 instance found!")
+        set(add_inst 1)
+    endif()
+    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"fp64\"" AND DTYPES MATCHES "fp64")
+        #message("fp64 instance found!")
+        set(add_inst 1)
+    endif()
+    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"bf16\"" AND DTYPES MATCHES "bf16")
+        #message("bf16 instance found!")
+        set(add_inst 1)
+    endif()
+    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"int8\"" AND DTYPES MATCHES "int8")
+        #message("int8 instance found!")
+        set(add_inst 1)
+    endif()
+    if(NOT "${cmake_instance}" MATCHES "DTYPES")
+        #message("instance should be built for all types!")
+        set(add_inst 1)
+    endif()
+    if(add_inst EQUAL 1 OR NOT DEFINED DTYPES)
+        list(APPEND CK_DEVICE_INSTANCES device_${subdir_path}_instance)
+    endif()
+ENDIF()
 ENDFOREACH()
+
 add_custom_target(instances DEPENDS utility;${CK_DEVICE_INSTANCES}  SOURCES ${INSTANCE_FILES})
+add_subdirectory(library)
 
-rocm_package_setup_component(tests
+if(NOT DEFINED INSTANCES_ONLY)
+ if(NOT DEFINED PROFILER_ONLY)
+   rocm_package_setup_component(tests
         LIBRARY_NAME composablekernel
         PACKAGE_NAME tests # Prevent -static suffix on package name
-)
+   )
 
-rocm_package_setup_component(examples
+   rocm_package_setup_component(examples
         LIBRARY_NAME composablekernel
         PACKAGE_NAME examples
-)
+   )
+   add_subdirectory(example)
+   add_subdirectory(test)
 
-rocm_package_setup_component(profiler
+   rocm_package_setup_component(profiler
         LIBRARY_NAME composablekernel
         PACKAGE_NAME ckProfiler
-)
-
-add_subdirectory(library)
-add_subdirectory(example)
-add_subdirectory(test)
-add_subdirectory(profiler)
+   )
+   add_subdirectory(profiler)
+  else()
+    #When building PROFILER_ONLY, label the package with GPU_ARCH
+    rocm_package_setup_component(profiler
+       LIBRARY_NAME composablekernel
+       PACKAGE_NAME ckProfiler_${GPU_ARCH}
+    )
+    add_subdirectory(profiler)
+  endif()
+endif()
 
 #Create an interface target for the include only files and call it "composablekernels"
 include(CMakePackageConfigHelpers)
 
-set(version 1.0.0)
 write_basic_package_version_file(
     "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfigVersion.cmake"
     VERSION "${version}"
@@ -295,9 +437,9 @@ write_basic_package_version_file(
 )
 
 configure_package_config_file(${CMAKE_CURRENT_SOURCE_DIR}/Config.cmake.in
-        "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfig.cmake"
-        INSTALL_DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
-        NO_CHECK_REQUIRED_COMPONENTS_MACRO
+    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfig.cmake"
+    INSTALL_DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
+    NO_CHECK_REQUIRED_COMPONENTS_MACRO
 )
 
 rocm_install(FILES
@@ -306,6 +448,13 @@ rocm_install(FILES
     DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
 )
 
+# Install CK version and configuration files
+rocm_install(FILES
+    ${PROJECT_BINARY_DIR}/include/ck/version.h
+    ${PROJECT_BINARY_DIR}/include/ck/config.h
+    DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/ck/
+)
+
 set(CPACK_RESOURCE_FILE_LICENSE "${CMAKE_CURRENT_SOURCE_DIR}/LICENSE")
 set(CPACK_RPM_PACKAGE_LICENSE "MIT")
 

CONTRIBUTORS.md

@@ -6,9 +6,11 @@ This is the list of developers and contributors to Composable Kernel library
 ## Developers
 [Chao Liu](https://github.com/asroy), [Jing Zhang](https://github.com/zjing14), 2018-2023
 
-[Letao Qin](https://github.com/ltqin), [Qianfeng Zhang](https://github.com/qianfengz), [Liang Huang](https://github.com/carlushuang), [Shaojie Wang](https://github.com/shaojiewang), 2019-2022
+[Letao Qin](https://github.com/ltqin), [Qianfeng Zhang](https://github.com/qianfengz), [Liang Huang](https://github.com/carlushuang), [Shaojie Wang](https://github.com/shaojiewang), 2019-2023
 
-[Anthony Chang](https://github.com/rosenrodt), [Chunyu Lai](https://github.com/rocking5566), [Illia Silin](https://github.com/illsilin), [Adam Osewski](https://github.com/aosewski), [Poyen Chen](https://github.com/poyenc), [Rosty Geyyer](https://github.com/geyyer), 2022
+[Anthony Chang](https://github.com/rosenrodt), [Chunyu Lai](https://github.com/rocking5566), [Illia Silin](https://github.com/illsilin), [Adam Osewski](https://github.com/aosewski), [Poyen Chen](https://github.com/poyenc), [Rosty Geyyer](https://github.com/geyyer), [Astha Rai](https://github.com/arai713), [Shi YanXing](https://github.com/Yanxing-Shi), 2022-2023
+
+[Hari Sadasivan](https://github.com/hsadasiv), [Bartlomiej Kocot](https://github.com/bartekxk), [Bartlomiej Wroblewski](https://github.com/bwroblew), 2023
 
 Hanwen Chang, 2019-2021,
 

Dockerfile

@@ -12,24 +12,32 @@ RUN useradd -rm -d /home/jenkins -s /bin/bash -u 1004 jenkins
 RUN chmod 1777 /tmp
 RUN apt-get update
 RUN apt-get install -y --allow-unauthenticated apt-utils wget gnupg2 curl
-RUN if [ "$ROCMVERSION" != "5.6" ]; then \
+
+ENV APT_KEY_DONT_WARN_ON_DANGEROUS_USAGE=DontWarn
+RUN curl -fsSL https://repo.radeon.com/rocm/rocm.gpg.key | gpg --dearmor -o /etc/apt/trusted.gpg.d/rocm-keyring.gpg
+
+RUN wget https://repo.radeon.com/amdgpu-install/5.6/ubuntu/focal/amdgpu-install_5.6.50600-1_all.deb  --no-check-certificate
+RUN apt-get update && \
+DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-unauthenticated \
+    ./amdgpu-install_5.6.50600-1_all.deb
+
+RUN if [ "$ROCMVERSION" != "5.7" ]; then \
         wget -qO - http://repo.radeon.com/rocm/rocm.gpg.key | apt-key add - && \
-        sh -c "echo deb [arch=amd64] $DEB_ROCM_REPO ubuntu main > /etc/apt/sources.list.d/rocm.list"; \
-    elif [ "$ROCMVERSION" = "5.6" ] && [ "$compiler_version" = "" ]; then \
-         sh -c "wget http://artifactory-cdn.amd.com/artifactory/list/amdgpu-deb/amd-nonfree-radeon_20.04-1_all.deb" && \
-         apt update && apt-get install -y ./amd-nonfree-radeon_20.04-1_all.deb && \
-         amdgpu-repo --amdgpu-build=1567752 --rocm-build=compute-rocm-dkms-no-npi-hipclang/11914 && \
-         amdgpu-install -y --usecase=rocm --no-dkms; \
-    elif [ "$ROCMVERSION" = "5.6" ] && [ "$compiler_version" = "rc3" ] || [ "$compiler_version" = "amd-stg-open" ]; then \
-         sh -c "wget http://artifactory-cdn.amd.com/artifactory/list/amdgpu-deb/amdgpu-install-internal_5.6-20.04-1_all.deb" && \
-         apt update && apt-get install -y ./amdgpu-install-internal_5.6-20.04-1_all.deb && \
-         sh -c 'echo deb [arch=amd64 trusted=yes] http://compute-artifactory.amd.com/artifactory/list/rocm-release-archive-20.04-deb/ 5.6 rel-45  > /etc/apt/sources.list.d/rocm-build.list' && \
-         amdgpu-repo --amdgpu-build=1602498 && amdgpu-install -y --usecase=rocm --no-dkms; \
+        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 [ "$ROCMVERSION" = "5.7" ] && [ "$compiler_version" = "" ] || [ "$compiler_version" = "amd-stg-open" ]; then \
+        sh -c "wget http://artifactory-cdn.amd.com/artifactory/list/amdgpu-deb/amdgpu-install-internal_5.7-20.04-1_all.deb" && \
+        apt update && apt-get install -y ./amdgpu-install-internal_5.7-20.04-1_all.deb && \
+        amdgpu-repo --amdgpu-build=1609671 --rocm-build=compute-rocm-npi-mi300/1354; \
+    elif [ "$ROCMVERSION" = "5.7" ] && [ "$compiler_version" = "rc1" ]; then \
+        sh -c "wget http://artifactory-cdn.amd.com/artifactory/list/amdgpu-deb/amdgpu-install-internal_5.7-20.04-1_all.deb" && \
+        apt update && apt-get install -y ./amdgpu-install-internal_5.7-20.04-1_all.deb && \
+        sh -c 'echo deb [arch=amd64 trusted=yes] http://compute-artifactory.amd.com/artifactory/list/rocm-release-archive-20.04-deb/ 5.7 rel-19 > /etc/apt/sources.list.d/rocm-build.list' && \
+        amdgpu-repo --amdgpu-build=1637781; \
     fi
 
-RUN wget --no-check-certificate -qO - https://apt.kitware.com/keys/kitware-archive-latest.asc 2>/dev/null | apt-key add -
 RUN sh -c "echo deb http://mirrors.kernel.org/ubuntu focal main universe | tee -a /etc/apt/sources.list"
-RUN curl -fsSL https://repo.radeon.com/rocm/rocm.gpg.key | gpg --dearmor -o /etc/apt/trusted.gpg.d/rocm-keyring.gpg
+RUN amdgpu-install -y --usecase=rocm --no-dkms
 
 # Install dependencies
 RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-unauthenticated \
@@ -45,6 +53,7 @@ RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-
     libpthread-stubs0-dev \
     llvm-amdgpu \
     pkg-config \
+    python \
     python3 \
     python3-dev \
     python3-pip \
@@ -54,12 +63,16 @@ RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-
     nano \
     zlib1g-dev \
     openssh-server \
-    clang-format-10 \
+    clang-format-12 \
     kmod && \
     apt-get clean && \
     rm -rf /var/lib/apt/lists/*
 
 #Install latest version of cmake
+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
 RUN apt purge --auto-remove -y cmake
 RUN apt update
 RUN apt install -y software-properties-common lsb-release

Jenkinsfile

@@ -11,6 +11,20 @@ def show_node_info() {
     """
 }
 
+def nthreads() {
+    def nproc = sh(returnStdout: true, script: 'nproc')
+    echo "Number of cores: ${nproc}"
+    def n = nproc.toInteger()
+    if (n > 32){
+        n /= 2
+    }
+    if (n > 64){
+        n = 64
+    }
+    echo "Number of threads used for building: ${n}"
+    return n
+}
+
 def runShell(String command){
     def responseCode = sh returnStatus: true, script: "${command} > tmp.txt"
     def output = readFile(file: "tmp.txt")
@@ -19,7 +33,7 @@ def runShell(String command){
 
 def getDockerImageName(){
     def img
-    if (params.ROCMVERSION != "5.6"){
+    if (params.ROCMVERSION != "5.7"){
        if (params.COMPILER_VERSION == "") {
            img = "${env.CK_DOCKERHUB}:ck_ub20.04_rocm${params.ROCMVERSION}"
        }
@@ -219,7 +233,8 @@ def cmake_build(Map conf=[:]){
         """
     def setup_cmd = conf.get("setup_cmd", "${cmake_envs} cmake ${setup_args}   .. ")
     // reduce parallelism when compiling, clang uses too much memory
-    def build_cmd = conf.get("build_cmd", "${build_envs} dumb-init make  -j\$(( \$(nproc) / 2 )) ${config_targets}")
+    def nt = nthreads()
+    def build_cmd = conf.get("build_cmd", "${build_envs} dumb-init make  -j${nt} ${config_targets}")
     def execute_cmd = conf.get("execute_cmd", "")
 
     def cmd = conf.get("cmd", """
@@ -461,7 +476,7 @@ def Build_CK(Map conf=[:]){
                         else{
                             echo "GPU is OK"
                         }
-                        if ( runShell('grep -n "gfx1030" clinfo.log') ){
+                        if ( runShell('grep -n "gfx1030" clinfo.log') || runShell('grep -n "gfx1101" clinfo.log') ){
                             navi_node = 1
                         }
                     }
@@ -482,7 +497,7 @@ def Build_CK(Map conf=[:]){
                         else{
                             echo "GPU is OK"
                         }
-                        if ( runShell('grep -n "gfx1030" clinfo.log') ){
+                        if ( runShell('grep -n "gfx1030" clinfo.log') || runShell('grep -n "gfx1101" clinfo.log') ){
                             navi_node = 1
                         }
                     }
@@ -493,8 +508,8 @@ def Build_CK(Map conf=[:]){
                 {
                     cmake_build(conf)
                     dir("build"){
-                        //run tests and examples 	
-                        sh 'make -j\$(( \$(nproc) / 2 )) check'
+                        //run tests and examples
+                        sh 'make -j check'
                         if (navi_node == 0 ){
                             //we only need the ckProfiler to run the performance tests, so we pack and stash it
                             //do not stash profiler on Navi nodes
@@ -597,8 +612,8 @@ def process_results(Map conf=[:]){
 }
 
 //launch develop branch daily at 23:00 UT in FULL_QA mode and at 19:00 UT with latest staging compiler version
-CRON_SETTINGS = BRANCH_NAME == "develop" ? '''0 23 * * * % RUN_FULL_QA=true
-                                              0 21 * * * % ROCMVERSION=5.5;COMPILER_VERSION=release;COMPILER_COMMIT=
+CRON_SETTINGS = BRANCH_NAME == "develop" ? '''0 23 * * * % RUN_FULL_QA=true;ROCMVERSION=5.7;COMPILER_VERSION=rc1
+                                              0 21 * * * % ROCMVERSION=5.6;COMPILER_VERSION=;COMPILER_COMMIT=
                                               0 19 * * * % BUILD_DOCKER=true;COMPILER_VERSION=amd-stg-open;COMPILER_COMMIT=''' : ""
 
 pipeline {
@@ -674,7 +689,7 @@ pipeline {
                                 -o -iname \'*.cpp.in\' \
                                 -o -iname \'*.cl\' \
                                 | grep -v 'build/' \
-                                | xargs -n 1 -P 1 -I{} -t sh -c \'clang-format-10 -style=file {} | diff - {}\'"
+                                | xargs -n 1 -P 1 -I{} -t sh -c \'clang-format-12 -style=file {} | diff - {}\'"
                     }
                     steps{
                         buildHipClangJobAndReboot(setup_cmd: "", build_cmd: "", execute_cmd: execute_cmd, no_reboot:true)
@@ -695,8 +710,8 @@ pipeline {
                     }
                     agent{ label rocmnode("gfx908 || gfx90a") }
                     environment{
-                        setup_args = """ -DCMAKE_INSTALL_PREFIX=../install -DGPU_TARGETS="gfx908;gfx90a;gfx940" """
-                        execute_args = """ cd ../client_example && rm -rf build && mkdir build && cd build && cmake -D CMAKE_PREFIX_PATH="${env.WORKSPACE}/install;/opt/rocm" -DGPU_TARGETS="gfx908;gfx90a;gfx940;gfx941;gfx942" -D CMAKE_CXX_COMPILER="${build_compiler()}" .. && make -j """ 
+                        setup_args = """ -DCMAKE_INSTALL_PREFIX=../install -DGPU_TARGETS="gfx908;gfx90a;gfx940;gfx941" """
+                        execute_args = """ cd ../client_example && rm -rf build && mkdir build && cd build && cmake -D CMAKE_PREFIX_PATH="${env.WORKSPACE}/install;/opt/rocm" -DGPU_TARGETS="gfx908;gfx90a;gfx940;gfx941" -D CMAKE_CXX_COMPILER="${build_compiler()}" .. && make -j """ 
                     }
                     steps{
                         Build_CK_and_Reboot(setup_args: setup_args, config_targets: "install", no_reboot:true, build_type: 'Release', execute_cmd: execute_args, prefixpath: '/usr/local')
@@ -717,7 +732,7 @@ pipeline {
                         Build_CK_and_Reboot(setup_args: setup_args, config_targets: "install", no_reboot:true, build_type: 'Release', execute_cmd: execute_args, prefixpath: '/usr/local')
                     }
                 }
-                stage("Build CK and run Tests on Navi")
+                stage("Build CK and run Tests on Navi21")
                 {
                     when {
                         beforeAgent true
@@ -725,7 +740,7 @@ pipeline {
                     }
                     agent{ label rocmnode("navi21") }
                     environment{
-                        setup_args = """ -DCMAKE_INSTALL_PREFIX=../install -DGPU_TARGETS="gfx1030" """ 
+                        setup_args = """ -DCMAKE_INSTALL_PREFIX=../install -DGPU_TARGETS="gfx1030" -DDL_KERNELS=ON """ 
                         execute_args = """ cd ../client_example && rm -rf build && mkdir build && cd build && cmake -D CMAKE_PREFIX_PATH="${env.WORKSPACE}/install;/opt/rocm" -DGPU_TARGETS="gfx1030" -D CMAKE_CXX_COMPILER="${build_compiler()}" .. && make -j """
 
                     }
@@ -733,6 +748,22 @@ pipeline {
                         Build_CK_and_Reboot(setup_args: setup_args, config_targets: "install", no_reboot:true, build_type: 'Release', execute_cmd: execute_args, prefixpath: '/usr/local')
                     }
                 }
+                stage("Build CK and run Tests on Navi32")
+                {
+                    when {
+                        beforeAgent true
+                        expression { !params.RUN_FULL_QA.toBoolean() }
+                    }
+                    agent{ label rocmnode("navi32") }
+                    environment{
+                        setup_args = """ -DCMAKE_INSTALL_PREFIX=../install -DDTYPES="fp16;fp32;bf16" -DGPU_TARGETS="gfx1101" """
+                        execute_args = """ cd ../client_example && rm -rf build && mkdir build && cd build && cmake -D CMAKE_PREFIX_PATH="${env.WORKSPACE}/install;/opt/rocm" -DGPU_TARGETS="gfx1101" -DDTYPES="fp16;fp32;bf16" -D CMAKE_CXX_COMPILER="${build_compiler()}" .. && make -j """
+
+                    }
+                    steps{
+                        Build_CK_and_Reboot(setup_args: setup_args, config_targets: "install", no_reboot:true, build_type: 'Release', execute_cmd: execute_args, prefixpath: '/usr/local')
+                    }
+                }
             }
         }
 

README.md

@@ -52,6 +52,8 @@ CK is released under the MIT license. [License File](/LICENSE)
 ```bash
 DOCKER_BUILDKIT=1 docker build -t ck:latest -f Dockerfile .
 ```
+Pre-built dockers are available from this public repo: 
+https://hub.docker.com/r/rocm/composable_kernel/tags
 
 ## Launch docker
 
@@ -76,12 +78,26 @@ mkdir build && cd build
 cmake                                                                                             \
 -D CMAKE_PREFIX_PATH=/opt/rocm                                                                    \
 -D CMAKE_CXX_COMPILER=/opt/rocm/bin/hipcc                                                         \
--D CMAKE_CXX_FLAGS="-O3"                                                                          \
 -D CMAKE_BUILD_TYPE=Release                                                                       \
 -D GPU_TARGETS="gfx908;gfx90a"                                                                    \
 ..
 ```
 
+If GPU_TARGETS is not set on the cmake command line, CK will be built for all targets supported by the 
+current compiler.
+
+
+Additional cmake flags can be used to significantly speed-up the build:
+
+INSTANCES_ONLY (by default is OFF) must be set to ON in order to build only the instances and library
+while skipping all tests, examples, and profiler. This is useful for libraries that use CK as a dependency.
+
+DTYPES (by default not set) can be set to any subset of "fp64;fp32;fp16;fp8;bf16;int8" to build instances 
+of select data types only. Currently, building of int8 instances is taking a lot of time (the compiler fix is in the works).
+
+DL_KERNELS (by default is OFF) must be set to ON in order to build the gemm_dl and batched_gemm_multi_d_dl 
+instances. Those instances are only needed for the NAVI2x platforms.
+
 ### Build examples and tests
 
 ```bash
@@ -109,6 +125,24 @@ make install
 
 Instructions for using CK as a pre-built kernel library are under [client_example](/client_example)
 
+## Contributing
+
+When you contribute to Composable Kernel, make sure to run `clang-format` on all the changed files. We highly recommend using git hooks that are managed by the `pre-commit` framework. To install hooks, run:
+
+```bash
+sudo script/install_precommit.sh
+```
+
+This way, `pre-commit` will add the appropriate hooks to your local repository and automatically run `clang-format` (and possibly additional checks) before any commit is created.
+
+If you need to uninstall hooks from the repository, you can do so by running the following command:
+
+```bash
+script/uninstall_precommit.sh
+```
+
+If for any reason, you need to temporarily disable precommit hooks, you can add the `--no-verify` option to the `git commit` command.
+
 ## Caveat
 ### Kernel Timing and Verification