biondizzle/nvfp4-megamoe-kernel
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Embed DeepGEMM source (not submodule) for SM100 raw CUDA GEMM primitives

Browse Source
This commit is contained in:
biondizzle
2026-06-01 07:39:40 +00:00
parent dae83723a3
commit e3ea609ddd
145 changed files with 27360 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
third_party/DeepGEMM/csrc/python_api.cpp vendored Normal file
View File

@@ -0,0 +1,28 @@
#include <pybind11/pybind11.h>
#include <torch/python.h>
#include "apis/attention.hpp"
#include "apis/einsum.hpp"
#include "apis/hyperconnection.hpp"
#include "apis/gemm.hpp"
#include "apis/layout.hpp"
#include "apis/mega.hpp"
#include "apis/runtime.hpp"
#ifndef TORCH_EXTENSION_NAME
#define TORCH_EXTENSION_NAME _C
#endif
// ReSharper disable once CppParameterMayBeConstPtrOrRef
PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) {
m.doc() = "DeepGEMM C++ library";
// TODO: make SM80 incompatible issues raise errors
deep_gemm::attention::register_apis(m);
deep_gemm::einsum::register_apis(m);
deep_gemm::hyperconnection::register_apis(m);
deep_gemm::gemm::register_apis(m);
deep_gemm::layout::register_apis(m);
deep_gemm::mega::register_apis(m);
deep_gemm::runtime::register_apis(m);
}