[PR #174] [MERGED] support compile with cuda 13.0 #278

New issue

Closed

opened 2026-10-26 21:15:42 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-26 21:15:42 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/deepseek-ai/DeepGEMM/pull/174
Author: @rainj-me
Created: 8/26/2025
Status: ✅ Merged
Merged: 8/27/2025
Merged by: @LyricZhao

Base: main ← Head: dev/support_cuda130

📝 Commits (1)

764ac98 support compile with cuda 13.0

📊 Changes

2 files changed (+2 additions, -1 deletions)

View changed files

📝 setup.py (+1 -0)
📝 third-party/cutlass (+1 -1)

📄 Description

Problem

When compile DeepGeMM with cuda 13.0, it throws the following error

creating build/temp.linux-x86_64-cpython-312/csrc
x86_64-linux-gnu-g++ -fno-strict-overflow -Wsign-compare -DNDEBUG -g -O2 -Wall -fPIC -I/usr/local/cuda/include -Ideep_gemm/include -Ithird-party/cutlass/include -Ithird-party/fmt/include -I/usr/local/lib/python3.12/dist-packages/torch/include -I/usr/local/lib/python3.12/dist-packages/torch/include/torch/csrc/api/include -I/usr/local/cuda/include -I/usr/include/python3.12 -c csrc/python_api.cpp -o build/temp.linux-x86_64-cpython-312/csrc/python_api.o -std=c++17 -O3 -fPIC -Wno-psabi -Wno-deprecated-declarations -D_GLIBCXX_USE_CXX11_ABI=1
In file included from third-party/cutlass/include/cutlass/platform/platform.h:101,
                 from third-party/cutlass/include/cutlass/arch/config.h:38,
                 from third-party/cutlass/include/cute/arch/config.hpp:33,
                 from third-party/cutlass/include/cute/arch/mma_sm100_desc.hpp:41,
                 from csrc/apis/../jit_kernels/impls/../heuristics/sm90.hpp:3,
                 from csrc/apis/../jit_kernels/impls/sm90_fp8_gemm_1d2d.hpp:10,
                 from csrc/apis/gemm.hpp:3,
                 from csrc/python_api.cpp:4:
third-party/cutlass/include/cutlass/cutlass.h:43:35: fatal error: cuda/std/utility: No such file or directory
   43 |   #define CUDA_STD_HEADER(header) <cuda/std/header>
      |                                   ^
compilation terminated.
error: command '/usr/bin/x86_64-linux-gnu-g++' failed with exit code 1

Description

This PR make DeepGemm successfully compile on cuda 13.0 and 12.8

Logs

root@ubuntu:/data01/rain/DeepGEMM_orig# pip list |grep cuda
cuda-bindings              13.0.0
cuda-pathfinder            1.0.0
cuda-python                13.0.0
cupy-cuda12x               13.3.0
dask-cuda                  25.6.0
numba-cuda                 0.11.0
nvidia-dali-cuda130        1.51.2
nvidia-resiliency-ext      0.4.1+cuda13
root@ubuntu:/data01/rain/DeepGEMM_orig# ./install.sh
running bdist_wheel
running build
running build_py
copying deep_gemm/__init__.py -> build/lib.linux-x86_64-cpython-312/deep_gemm
creating build/lib.linux-x86_64-cpython-312/deep_gemm/testing
copying deep_gemm/testing/numeric.py -> build/lib.linux-x86_64-cpython-312/deep_gemm/testing
...
creating build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/common
copying deep_gemm/include/deep_gemm/common/sm100_utils.cuh -> build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/common
...
copying deep_gemm/include/deep_gemm/common/cute_tie.cuh -> build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/common
creating build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/impls
copying deep_gemm/include/deep_gemm/impls/sm100_fp8_gemm_1d1d.cuh -> build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/impls
...
copying deep_gemm/include/deep_gemm/impls/sm90_bf16_gemm.cuh -> build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/impls
running build_ext
building 'deep_gemm_cpp' extension
creating build/temp.linux-x86_64-cpython-312/csrc
x86_64-linux-gnu-g++ -fno-strict-overflow -Wsign-compare -DNDEBUG -g -O2 -Wall -fPIC -I/usr/local/cuda/include -I/usr/local/cuda/include/cccl -Ideep_gemm/include -Ithird-party/cutlass/include -Ithird-party/fmt/include -I/usr/local/lib/python3.12/dist-packages/torch/include -I/usr/local/lib/python3.12/dist-packages/torch/include/torch/csrc/api/include -I/usr/local/cuda/include -I/usr/include/python3.12 -c csrc/python_api.cpp -o build/temp.linux-x86_64-cpython-312/csrc/python_api.o -std=c++17 -O3 -fPIC -Wno-psabi -Wno-deprecated-declarations -D_GLIBCXX_USE_CXX11_ABI=1
x86_64-linux-gnu-g++ -fno-strict-overflow -Wsign-compare -DNDEBUG -g -O2 -Wall -shared -Wl,-O1 -Wl,-Bsymbolic-functions build/temp.linux-x86_64-cpython-312/csrc/python_api.o -L/usr/local/cuda/lib64 -L/usr/local/cuda/lib64/stubs -L/usr/local/lib/python3.12/dist-packages/torch/lib -L/usr/local/cuda/lib64 -L/usr/lib/x86_64-linux-gnu -lcuda -lcudart -lnvrtc -lc10 -ltorch -ltorch_cpu -ltorch_python -lcudart -lc10_cuda -ltorch_cuda -o build/lib.linux-x86_64-cpython-312/deep_gemm_cpp.cpython-312-x86_64-linux-gnu.so
/usr/local/lib/python3.12/dist-packages/setuptools/_distutils/cmd.py:90: SetuptoolsDeprecationWarning: setup.py install is deprecated.
!!

        ********************************************************************************
        Please avoid running ``setup.py`` directly.
        Instead, use pypa/build, pypa/installer or other
        standards-based tools.

        See https://blog.ganssle.io/articles/2021/10/setup-py-deprecated.html for details.
        ********************************************************************************

!!
  self.initialize_options()
installing to build/bdist.linux-x86_64/wheel
running install
running install_lib
creating build/bdist.linux-x86_64/wheel
copying build/lib.linux-x86_64-cpython-312/deep_gemm_cpp.cpython-312-x86_64-linux-gnu.so -> build/bdist.linux-x86_64/wheel/.
creating build/bdist.linux-x86_64/wheel/deep_gemm
creating build/bdist.linux-x86_64/wheel/deep_gemm/testing
copying build/lib.linux-x86_64-cpython-312/deep_gemm/testing/numeric.py -> build/bdist.linux-x86_64/wheel/./deep_gemm/testing
...
copying build/lib.linux-x86_64-cpython-312/deep_gemm/utils/layout.py -> build/bdist.linux-x86_64/wheel/./deep_gemm/utils
running install_egg_info
running egg_info
creating deep_gemm.egg-info
writing deep_gemm.egg-info/PKG-INFO
writing dependency_links to deep_gemm.egg-info/dependency_links.txt
writing top-level names to deep_gemm.egg-info/top_level.txt
writing manifest file 'deep_gemm.egg-info/SOURCES.txt'
reading manifest file 'deep_gemm.egg-info/SOURCES.txt'
adding license file 'LICENSE'
writing manifest file 'deep_gemm.egg-info/SOURCES.txt'
Copying deep_gemm.egg-info to build/bdist.linux-x86_64/wheel/./deep_gemm-2.0.0+453e771-py3.12.egg-info
running install_scripts
creating build/bdist.linux-x86_64/wheel/deep_gemm-2.0.0+453e771.dist-info/WHEEL
creating 'dist/deep_gemm-2.0.0+453e771-cp312-cp312-linux_x86_64.whl' and adding 'build/bdist.linux-x86_64/wheel' to it
adding 'deep_gemm_cpp.cpython-312-x86_64-linux-gnu.so'
adding 'deep_gemm/__init__.py'
...
adding 'deep_gemm-2.0.0+453e771.dist-info/RECORD'
removing build/bdist.linux-x86_64/wheel
Processing ./dist/deep_gemm-2.0.0+453e771-cp312-cp312-linux_x86_64.whl
Installing collected packages: deep-gemm
  Attempting uninstall: deep-gemm
    Found existing installation: deep_gemm 2.0.0+9fe2798
    Uninstalling deep_gemm-2.0.0+9fe2798:
      Successfully uninstalled deep_gemm-2.0.0+9fe2798
Successfully installed deep-gemm-2.0.0+453e771

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/deepseek-ai/DeepGEMM/pull/174 **Author:** [@rainj-me](https://github.com/rainj-me) **Created:** 8/26/2025 **Status:** ✅ Merged **Merged:** 8/27/2025 **Merged by:** [@LyricZhao](https://github.com/LyricZhao) **Base:** `main` ← **Head:** `dev/support_cuda130` --- ### 📝 Commits (1) - [`764ac98`](https://github.com/deepseek-ai/DeepGEMM/commit/764ac9816b858f18d477df847a50a1d37e72445a) support compile with cuda 13.0 ### 📊 Changes **2 files changed** (+2 additions, -1 deletions) <details> <summary>View changed files</summary> 📝 `setup.py` (+1 -0) 📝 `third-party/cutlass` (+1 -1) </details> ### 📄 Description ## Problem When compile DeepGeMM with cuda 13.0, it throws the following error ``` creating build/temp.linux-x86_64-cpython-312/csrc x86_64-linux-gnu-g++ -fno-strict-overflow -Wsign-compare -DNDEBUG -g -O2 -Wall -fPIC -I/usr/local/cuda/include -Ideep_gemm/include -Ithird-party/cutlass/include -Ithird-party/fmt/include -I/usr/local/lib/python3.12/dist-packages/torch/include -I/usr/local/lib/python3.12/dist-packages/torch/include/torch/csrc/api/include -I/usr/local/cuda/include -I/usr/include/python3.12 -c csrc/python_api.cpp -o build/temp.linux-x86_64-cpython-312/csrc/python_api.o -std=c++17 -O3 -fPIC -Wno-psabi -Wno-deprecated-declarations -D_GLIBCXX_USE_CXX11_ABI=1 In file included from third-party/cutlass/include/cutlass/platform/platform.h:101, from third-party/cutlass/include/cutlass/arch/config.h:38, from third-party/cutlass/include/cute/arch/config.hpp:33, from third-party/cutlass/include/cute/arch/mma_sm100_desc.hpp:41, from csrc/apis/../jit_kernels/impls/../heuristics/sm90.hpp:3, from csrc/apis/../jit_kernels/impls/sm90_fp8_gemm_1d2d.hpp:10, from csrc/apis/gemm.hpp:3, from csrc/python_api.cpp:4: third-party/cutlass/include/cutlass/cutlass.h:43:35: fatal error: cuda/std/utility: No such file or directory 43 | #define CUDA_STD_HEADER(header) <cuda/std/header> | ^ compilation terminated. error: command '/usr/bin/x86_64-linux-gnu-g++' failed with exit code 1 ``` ## Description This PR make DeepGemm successfully compile on cuda 13.0 and 12.8 ## Logs ``` root@ubuntu:/data01/rain/DeepGEMM_orig# pip list |grep cuda cuda-bindings 13.0.0 cuda-pathfinder 1.0.0 cuda-python 13.0.0 cupy-cuda12x 13.3.0 dask-cuda 25.6.0 numba-cuda 0.11.0 nvidia-dali-cuda130 1.51.2 nvidia-resiliency-ext 0.4.1+cuda13 root@ubuntu:/data01/rain/DeepGEMM_orig# ./install.sh running bdist_wheel running build running build_py copying deep_gemm/__init__.py -> build/lib.linux-x86_64-cpython-312/deep_gemm creating build/lib.linux-x86_64-cpython-312/deep_gemm/testing copying deep_gemm/testing/numeric.py -> build/lib.linux-x86_64-cpython-312/deep_gemm/testing ... creating build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/common copying deep_gemm/include/deep_gemm/common/sm100_utils.cuh -> build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/common ... copying deep_gemm/include/deep_gemm/common/cute_tie.cuh -> build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/common creating build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/impls copying deep_gemm/include/deep_gemm/impls/sm100_fp8_gemm_1d1d.cuh -> build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/impls ... copying deep_gemm/include/deep_gemm/impls/sm90_bf16_gemm.cuh -> build/lib.linux-x86_64-cpython-312/deep_gemm/include/deep_gemm/impls running build_ext building 'deep_gemm_cpp' extension creating build/temp.linux-x86_64-cpython-312/csrc x86_64-linux-gnu-g++ -fno-strict-overflow -Wsign-compare -DNDEBUG -g -O2 -Wall -fPIC -I/usr/local/cuda/include -I/usr/local/cuda/include/cccl -Ideep_gemm/include -Ithird-party/cutlass/include -Ithird-party/fmt/include -I/usr/local/lib/python3.12/dist-packages/torch/include -I/usr/local/lib/python3.12/dist-packages/torch/include/torch/csrc/api/include -I/usr/local/cuda/include -I/usr/include/python3.12 -c csrc/python_api.cpp -o build/temp.linux-x86_64-cpython-312/csrc/python_api.o -std=c++17 -O3 -fPIC -Wno-psabi -Wno-deprecated-declarations -D_GLIBCXX_USE_CXX11_ABI=1 x86_64-linux-gnu-g++ -fno-strict-overflow -Wsign-compare -DNDEBUG -g -O2 -Wall -shared -Wl,-O1 -Wl,-Bsymbolic-functions build/temp.linux-x86_64-cpython-312/csrc/python_api.o -L/usr/local/cuda/lib64 -L/usr/local/cuda/lib64/stubs -L/usr/local/lib/python3.12/dist-packages/torch/lib -L/usr/local/cuda/lib64 -L/usr/lib/x86_64-linux-gnu -lcuda -lcudart -lnvrtc -lc10 -ltorch -ltorch_cpu -ltorch_python -lcudart -lc10_cuda -ltorch_cuda -o build/lib.linux-x86_64-cpython-312/deep_gemm_cpp.cpython-312-x86_64-linux-gnu.so /usr/local/lib/python3.12/dist-packages/setuptools/_distutils/cmd.py:90: SetuptoolsDeprecationWarning: setup.py install is deprecated. !! ******************************************************************************** Please avoid running ``setup.py`` directly. Instead, use pypa/build, pypa/installer or other standards-based tools. See https://blog.ganssle.io/articles/2021/10/setup-py-deprecated.html for details. ******************************************************************************** !! self.initialize_options() installing to build/bdist.linux-x86_64/wheel running install running install_lib creating build/bdist.linux-x86_64/wheel copying build/lib.linux-x86_64-cpython-312/deep_gemm_cpp.cpython-312-x86_64-linux-gnu.so -> build/bdist.linux-x86_64/wheel/. creating build/bdist.linux-x86_64/wheel/deep_gemm creating build/bdist.linux-x86_64/wheel/deep_gemm/testing copying build/lib.linux-x86_64-cpython-312/deep_gemm/testing/numeric.py -> build/bdist.linux-x86_64/wheel/./deep_gemm/testing ... copying build/lib.linux-x86_64-cpython-312/deep_gemm/utils/layout.py -> build/bdist.linux-x86_64/wheel/./deep_gemm/utils running install_egg_info running egg_info creating deep_gemm.egg-info writing deep_gemm.egg-info/PKG-INFO writing dependency_links to deep_gemm.egg-info/dependency_links.txt writing top-level names to deep_gemm.egg-info/top_level.txt writing manifest file 'deep_gemm.egg-info/SOURCES.txt' reading manifest file 'deep_gemm.egg-info/SOURCES.txt' adding license file 'LICENSE' writing manifest file 'deep_gemm.egg-info/SOURCES.txt' Copying deep_gemm.egg-info to build/bdist.linux-x86_64/wheel/./deep_gemm-2.0.0+453e771-py3.12.egg-info running install_scripts creating build/bdist.linux-x86_64/wheel/deep_gemm-2.0.0+453e771.dist-info/WHEEL creating 'dist/deep_gemm-2.0.0+453e771-cp312-cp312-linux_x86_64.whl' and adding 'build/bdist.linux-x86_64/wheel' to it adding 'deep_gemm_cpp.cpython-312-x86_64-linux-gnu.so' adding 'deep_gemm/__init__.py' ... adding 'deep_gemm-2.0.0+453e771.dist-info/RECORD' removing build/bdist.linux-x86_64/wheel Processing ./dist/deep_gemm-2.0.0+453e771-cp312-cp312-linux_x86_64.whl Installing collected packages: deep-gemm Attempting uninstall: deep-gemm Found existing installation: deep_gemm 2.0.0+9fe2798 Uninstalling deep_gemm-2.0.0+9fe2798: Successfully uninstalled deep_gemm-2.0.0+9fe2798 Successfully installed deep-gemm-2.0.0+453e771 ``` --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-26 21:15:42 +00:00