Alternative for CUDAExtension for intel GPU(xpu)

yash3056 · March 3, 2025, 3:36pm

torch/utils/cpp_extension.py

1eba9b3aa


      
                  libraries.append('torch_python')
              if IS_WINDOWS and platform.machine().lower() != "arm64":
                  libraries.append("sleef")
          
              kwargs['libraries'] = libraries
          
              kwargs['language'] = 'c++'
              return setuptools.Extension(name, sources, *args, **kwargs)
          
          
          def CUDAExtension(name, sources, *args, **kwargs):
              """
              Create a :class:`setuptools.Extension` for CUDA/C++.
          
              Convenience method that creates a :class:`setuptools.Extension` with the
              bare minimum (but often sufficient) arguments to build a CUDA/C++
              extension. This includes the CUDA include path, library path and runtime
              library.
          
              All arguments are forwarded to the :class:`setuptools.Extension`
              constructor. Full list arguments can be found at

I am trying to utilize FlashMLA for intel GPUs, I need some help with the dependencies.