v4.2 release. (#2587)

* Fix default cluster callback values to 1 to avoid profiler failure when these values are not set in command line.

* v4.2 release.
This commit is contained in:
Junkai-Wu
2025-08-22 18:11:24 -04:00
committed by GitHub
parent 11cad1f67b
commit a49a78ffef
351 changed files with 28182 additions and 2032 deletions
+53 -6
View File
@@ -175,13 +175,25 @@ if (CUDA_VERSION VERSION_GREATER_EQUAL 12.0)
endif()
if (CUDA_VERSION VERSION_GREATER_EQUAL 12.8)
list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 100 100a 120 120a)
list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 101 101a)
list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 100 100a 120 120a 121 121a)
if (CUDA_VERSION VERSION_LESS 13.0)
list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 101 101a)
else()
list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 110 110a)
endif()
endif()
if (CUDA_VERSION VERSION_GREATER_EQUAL 12.9)
list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 100f 120f)
list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 101f)
list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 100f 120f 121f 103a 103f)
if (CUDA_VERSION VERSION_LESS 13.0)
list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 101f)
else()
list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 110f)
endif()
endif()
if (CUDA_VERSION VERSION_GREATER_EQUAL 13.0)
list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 110 110a)
endif()
set(CUTLASS_NVCC_ARCHS ${CUTLASS_NVCC_ARCHS_SUPPORTED} CACHE STRING "The SM architectures requested.")
@@ -288,18 +300,50 @@ if (KERNEL_FILTER_FILE)
set(KERNEL_FILTER_FILE "${KERNEL_FILTER_FILE}" CACHE STRING "KERNEL FILTER FILE FULL PATH" FORCE)
endif()
if (CUTLASS_LIBRARY_HEURISTICS_PROBLEMS_FILE)
get_filename_component(CUTLASS_LIBRARY_HEURISTICS_PROBLEMS_FILE "${CUTLASS_LIBRARY_HEURISTICS_PROBLEMS_FILE}" ABSOLUTE)
set(CUTLASS_LIBRARY_HEURISTICS_PROBLEMS_FILE "${CUTLASS_LIBRARY_HEURISTICS_PROBLEMS_FILE}" CACHE STRING "HEURISTICS FILE FULL PATH" FORCE)
endif()
set(SELECTED_KERNEL_LIST "selected" CACHE STRING "Name of the filtered kernel list")
if(KERNEL_FILTER_FILE)
message(STATUS "Full path of filter file: ${KERNEL_FILTER_FILE}")
endif()
if(CUTLASS_LIBRARY_HEURISTICS_PROBLEMS_FILE)
message(STATUS "Full path of heuristics problems file: ${CUTLASS_LIBRARY_HEURISTICS_PROBLEMS_FILE}")
if(DEFINED CUTLASS_NVMMH_URL)
message(STATUS "CUTLASS_NVVMH_URL is set. Fetching dependency")
include(FetchContent)
FetchContent_Declare(
nvmmh
URL ${CUTLASS_NVMMH_URL}
)
FetchContent_MakeAvailable(nvmmh)
FetchContent_GetProperties(nvmmh SOURCE_DIR nvmmh_dir)
set(CUTLASS_NVMMH_PATH "${nvmmh_dir}")
endif()
if(DEFINED CUTLASS_NVMMH_PATH)
message(STATUS "CUTLASS_NVMMH_PATH is set. Using package at: ${CUTLASS_NVMMH_PATH}")
set(CUTLASS_NVMMH_PY_DIR "${CUTLASS_NVMMH_PATH}/python/")
set(ENV{CUTLASS_NVMMH_SO_PATH} "${CUTLASS_NVMMH_PATH}/lib/libnvMatmulHeuristics.so")
endif()
endif()
set(CUTLASS_LIBRARY_OPERATIONS "all" CACHE STRING "Comma-delimited list of operation name filters. Default '' means all operations are enabled.")
set(CUTLASS_LIBRARY_KERNELS ${CUTLASS_LIBRARY_KERNELS_INIT} CACHE STRING "Comma-delimited list of kernel name filters. If unspecified, only the largest tile size is enabled. If the string 'all' is specified, all kernels are enabled.")
set(CUTLASS_LIBRARY_IGNORE_KERNELS "" CACHE STRING "Comma-delimited list of kernels to exclude from build. This option ONLY takes effect if CUTLASS_LIBRARY_KERNELS is set.")
set(CUTLASS_LIBRARY_EXCLUDE_KERNELS "" CACHE STRING "Comma-delimited list of kernels to exclude from build. This option always takes effect, whether or not CUTLASS_LIBRARY_KERNELS is set. It also can exclude kernels from the filter file (see KERNEL_FILTER_FILE).")
set(CUTLASS_LIBRARY_INSTANTIATION_LEVEL "" CACHE STRING "Instantiation level for SM90 kernels. Set to `max` and make sure CUTLASS_LIBRARY_KERNELS is non-empty to stamp all possible kernel configurations.")
if(CUTLASS_LIBRARY_INSTANTIATION_LEVEL OR CUTLASS_LIBRARY_HEURISTICS_PROBLEMS_FILE)
message(STATUS "Enable extended SM90 WGMMA instruction shapes for instantiation levels")
list(APPEND CUTLASS_CUDA_NVCC_FLAGS -DCUTE_SM90_EXTENDED_MMA_SHAPES_ENABLED)
endif()
################################################################################
set(CUTLASS_TEST_ENABLE_CACHED_RESULTS ON CACHE BOOL "Enable caching and reuse of test results in unit tests")
@@ -350,6 +394,10 @@ if (CUTLASS_NVCC_ARCHS MATCHES 100f OR CUTLASS_NVCC_ARCHS MATCHES 101f)
list(APPEND CUTLASS_CUDA_NVCC_FLAGS -DCUTLASS_SM100_FAMILY_ARCHS_ENABLED)
endif()
if (CUTLASS_NVCC_ARCHS MATCHES 110f)
list(APPEND CUTLASS_CUDA_NVCC_FLAGS -DCUTLASS_SM100_FAMILY_ARCHS_ENABLED)
endif()
set(CUTLASS_SKIP_REDUCTION_INIT OFF CACHE BOOL "Disable init reduction workspace")
#
@@ -428,8 +476,6 @@ endif()
#
###################################################################################################
# Warnings-as-error exceptions and warning suppressions for Clang builds
if (CUTLASS_CLANG_HOST_COMPILE)
@@ -705,6 +751,7 @@ target_include_directories(
SYSTEM INTERFACE
$<BUILD_INTERFACE:${CUDA_TOOLKIT_ROOT_DIR}/include>
)
if(CUDA_VERSION VERSION_GREATER_EQUAL 13.0)
target_include_directories(
CUTLASS