From fb7c687548b01ca53ff93d1ce1177b46b7303470 Mon Sep 17 00:00:00 2001 From: dan_the_3rd <43445237+danthe3rd@users.noreply.github.com> Date: Tue, 29 Jul 2025 10:14:15 +0200 Subject: [PATCH] Merge pull request #135 from danthe3rd/patch-3 Fix import-time error: "undefined symbol: _ZN3c104cuda20getCurrentCUDAStreamEa" --- setup.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/setup.py b/setup.py index 916f760..22a65ba 100644 --- a/setup.py +++ b/setup.py @@ -4,7 +4,7 @@ import shutil import subprocess from setuptools import find_packages from setuptools.command.build_py import build_py -from torch.utils.cpp_extension import CppExtension, CUDA_HOME +from torch.utils.cpp_extension import CUDAExtension, CUDA_HOME current_dir = os.path.dirname(os.path.realpath(__file__)) cxx_flags = ['-std=c++20', '-O3', '-fPIC', '-Wno-psabi'] @@ -89,7 +89,7 @@ if __name__ == '__main__': ] }, ext_modules=[ - CppExtension(name='deep_gemm_cpp', + CUDAExtension(name='deep_gemm_cpp', sources=sources, include_dirs=build_include_dirs, libraries=build_libraries,