# Copyright (c) 2024 - 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: BSD-3-Clause # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions are met: # # 1. Redistributions of source code must retain the above copyright notice, this # list of conditions and the following disclaimer. # # 2. Redistributions in binary form must reproduce the above copyright notice, # this list of conditions and the following disclaimer in the documentation # and/or other materials provided with the distribution. # # 3. Neither the name of the copyright holder nor the names of its # contributors may be used to endorse or promote products derived from # this software without specific prior written permission. # # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" # AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE # IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE # DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE # FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL # DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR # SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER # CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, # OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. # The purpose of this target is to check if the following header files are self-contained, # i.e. they can be included in a source file without needing to include other headers before it. set(header_files_to_check # cutlass # cutlass/gemm/kernel cutlass/gemm/kernel/default_gemm.h cutlass/gemm/kernel/default_gemm_complex.h cutlass/gemm/kernel/gemm_universal_decl.h # cutlass/gemm/kernel/sm90_gemm_warpspecialized.hpp # cute cute/config.hpp cute/int_tuple.hpp cute/layout.hpp cute/layout_composed.hpp cute/pointer.hpp cute/pointer_base.hpp cute/pointer_flagged.hpp cute/pointer_swizzle.hpp cute/stride.hpp cute/swizzle.hpp cute/swizzle_layout.hpp cute/tensor.hpp cute/tensor_impl.hpp cute/underscore.hpp # cute/algorithm cute/algorithm/axpby.hpp cute/algorithm/clear.hpp # cute/algorithm/cooperative_copy.hpp cute/algorithm/cooperative_gemm.hpp # cute/algorithm/copy.hpp cute/algorithm/fill.hpp cute/algorithm/functional.hpp # cute/algorithm/gemm.hpp cute/algorithm/prefer.hpp # cute/algorithm/prefetch.hpp cute/algorithm/tensor_algorithms.hpp cute/algorithm/tuple_algorithms.hpp # cute/container cute/container/alignment.hpp cute/container/array.hpp cute/container/array_aligned.hpp cute/container/array_subbyte.hpp cute/container/bit_field.hpp cute/container/cuda_types.hpp cute/container/tuple.hpp cute/container/type_list.hpp # cute/numeric cute/numeric/arithmetic_tuple.hpp cute/numeric/complex.hpp cute/numeric/int.hpp cute/numeric/integer_sequence.hpp cute/numeric/integral_ratio.hpp cute/numeric/math.hpp cute/numeric/numeric_types.hpp cute/numeric/real.hpp cute/numeric/integral_constant.hpp # cute/util cute/util/debug.hpp cute/util/print.hpp cute/util/type_traits.hpp # cute/arch cute/arch/cluster_sm90.hpp cute/arch/copy.hpp cute/arch/copy_sm50.hpp cute/arch/copy_sm75.hpp cute/arch/copy_sm80.hpp cute/arch/copy_sm90.hpp cute/arch/copy_sm90_desc.hpp cute/arch/copy_sm90_tma.hpp cute/arch/mma_sm61.hpp cute/arch/mma_sm70.hpp cute/arch/mma_sm75.hpp cute/arch/mma_sm80.hpp cute/arch/mma_sm90.hpp cute/arch/mma_sm90_desc.hpp cute/arch/mma_sm90_gmma.hpp cute/arch/mma.hpp cute/arch/util.hpp cute/arch/cluster_sm100.hpp cute/arch/copy_sm100.hpp cute/arch/copy_sm100_tma.hpp cute/arch/mma_sm100.hpp cute/arch/mma_sm100_desc.hpp cute/arch/mma_sm100_umma.hpp # cute/arch/tmem_allocator_sm100.hpp # cute/atom # cute/atom/copy_atom.hpp # cute/atom/copy_traits.hpp # cute/atom/copy_traits_sm50.hpp # cute/atom/copy_traits_sm75.hpp # cute/atom/copy_traits_sm80.hpp # cute/atom/copy_traits_sm90.hpp # cute/atom/copy_traits_sm90_im2col.hpp # cute/atom/copy_traits_sm90_tma.hpp # cute/atom/copy_traits_sm90_tma_swizzle.hpp cute/atom/mma_atom.hpp cute/atom/mma_traits.hpp cute/atom/mma_traits_sm61.hpp cute/atom/mma_traits_sm70.hpp cute/atom/mma_traits_sm75.hpp cute/atom/mma_traits_sm80.hpp cute/atom/mma_traits_sm90.hpp cute/atom/mma_traits_sm90_gmma.hpp cute/atom/mma_traits_sm100.hpp cute/atom/partitioner.hpp # cutlass cutlass/aligned_buffer.h cutlass/array.h cutlass/array_planar_complex.h cutlass/array_subbyte.h cutlass/barrier.h cutlass/bfloat16.h cutlass/blas3.h cutlass/blas3_types.h cutlass/block_striped.h cutlass/cluster_launch.hpp cutlass/complex.h cutlass/constants.h cutlass/coord.h cutlass/core_io.h cutlass/cuda_host_adapter.hpp cutlass/cutlass.h cutlass/device_kernel.h cutlass/fast_math.h cutlass/float8.h # cutlass/floating_point_nvrtc.h cutlass/functional.h cutlass/gemm_coord.h cutlass/gemm_coord.hpp cutlass/half.h cutlass/integer_subbyte.h cutlass/kernel_hardware_info.h cutlass/kernel_hardware_info.hpp cutlass/kernel_launch.h cutlass/matrix.h cutlass/matrix_coord.h cutlass/matrix_shape.h cutlass/numeric_conversion.h cutlass/numeric_size.h cutlass/numeric_types.h cutlass/pitch_linear_coord.h cutlass/predicate_vector.h cutlass/quaternion.h cutlass/real.h cutlass/relatively_equal.h cutlass/semaphore.h cutlass/subbyte_reference.h cutlass/tensor_coord.h cutlass/tensor_ref.h cutlass/tensor_ref_planar_complex.h cutlass/tensor_view.h cutlass/tensor_view_planar_complex.h cutlass/tfloat32.h cutlass/trace.h cutlass/uint128.h cutlass/version.h cutlass/wmma_array.h cutlass/workspace.h cutlass/exmy_base.h cutlass/float_subbyte.h # cutlass/platform cutlass/platform/platform.h # cutlass/pipeline cutlass/pipeline/pipeline.hpp cutlass/pipeline/sm90_pipeline.hpp cutlass/pipeline/sm100_pipeline.hpp # cutlass/detail cutlass/detail/cluster.hpp cutlass/detail/collective.hpp cutlass/detail/dependent_false.hpp cutlass/detail/helper_macros.hpp cutlass/detail/layout.hpp cutlass/detail/mma.hpp cutlass/detail/sm100_blockscaled_layout.hpp # cutlass/arch cutlass/arch/arch.h cutlass/arch/barrier.h cutlass/arch/cache_operation.h cutlass/arch/config.h cutlass/arch/grid_dependency_control.h cutlass/arch/memory.h # cutlass/arch/memory_sm75.h # cutlass/arch/memory_sm80.h cutlass/arch/mma.h # cutlass/arch/mma_sm50.h # cutlass/arch/mma_sm60.h # cutlass/arch/mma_sm61.h # cutlass/arch/mma_sm70.h # cutlass/arch/mma_sm75.h # cutlass/arch/mma_sm80.h # cutlass/arch/mma_sm89.h # cutlass/arch/mma_sm90.h cutlass/arch/mma_sparse_sm80.h cutlass/arch/mma_sparse_sm89.h # cutlass/arch/simd.h # cutlass/arch/simd_sm60.h # cutlass/arch/simd_sm61.h cutlass/arch/reg_reconfig.h cutlass/arch/wmma.h # cutlass/arch/wmma_sm70.h # cutlass/arch/wmma_sm72.h # cutlass/arch/wmma_sm75.h # cutlass/arch/wmma_sm80.h # cutlass/layout cutlass/layout/layout.h cutlass/layout/matrix.h cutlass/layout/permute.h cutlass/layout/pitch_linear.h cutlass/layout/tensor.h cutlass/layout/tensor_op_multiplicand_sm70.h cutlass/layout/tensor_op_multiplicand_sm75.h cutlass/layout/tensor_op_multiplicand_sm80.h cutlass/layout/vector.h ) # for each header in _header_files: # create a .cu file with the same name as the header's path, except with / replaced with % # have the .cu file include that header set(_gen_source_files "") foreach(header_file ${header_files_to_check}) string(REPLACE "/" "%" header_file_esc ${header_file}) file(WRITE "${CMAKE_CURRENT_BINARY_DIR}/${header_file_esc}.cu" "#include <${header_file}>") list(APPEND _gen_source_files "${CMAKE_CURRENT_BINARY_DIR}/${header_file_esc}.cu") endforeach() # build all generated .cu files into a single library cutlass_add_library(test_self_contained_includes MODULE ${_gen_source_files})