[CI/Build] Per file CUDA Archs (improve wheel size and dev build times) (#8845)

This commit is contained in:
Lucas Wilkinson
2024-10-03 22:55:25 -04:00
committed by GitHub
parent 3dbb215b38
commit aeb37c2a72
22 changed files with 828 additions and 370 deletions
@@ -26,6 +26,7 @@
#include <iostream>
#include "common/base.h"
#include "core/registration.h"
#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 800
#include "common/mem.h"
@@ -1066,3 +1067,7 @@ torch::Tensor marlin_gemm(torch::Tensor& a, torch::Tensor& b_q_weight,
return c;
}
TORCH_LIBRARY_IMPL_EXPAND(TORCH_EXTENSION_NAME, CUDA, m) {
m.impl("marlin_gemm", &marlin_gemm);
}
@@ -30,6 +30,7 @@
#include <iostream>
#include "../dense/common/base.h"
#include "core/registration.h"
#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 800
#include "../dense/common/mem.h"
@@ -1241,3 +1242,7 @@ torch::Tensor marlin_qqq_gemm(torch::Tensor const& a,
return d;
}
TORCH_LIBRARY_IMPL_EXPAND(TORCH_EXTENSION_NAME, CUDA, m) {
m.impl("marlin_qqq_gemm", &marlin_qqq_gemm);
}
@@ -28,6 +28,7 @@
#include "common/base.h"
#include "core/scalar_type.hpp"
#include "core/registration.h"
#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ < 800
@@ -1134,3 +1135,7 @@ torch::Tensor gptq_marlin_24_gemm(torch::Tensor& a, torch::Tensor& b_q_weight,
return c;
}
TORCH_LIBRARY_IMPL_EXPAND(TORCH_EXTENSION_NAME, CUDA, m) {
m.impl("gptq_marlin_24_gemm", &gptq_marlin_24_gemm);
}