7d1c661381
This is a patch in preparation for the support stream ordered memory allocator in CUDA Fortran. This patch adds an asynchronous id to the AllocatableAllocate runtime function and to Descriptor::Allocate so it can be passed down to the registered allocator. It is up to the allocator to use this value or not. A follow up patch will implement that asynchronous allocator for CUDA Fortran.
52 lines
1.6 KiB
C++
52 lines
1.6 KiB
C++
//===-- runtime/CUDA/descriptor.cpp ---------------------------------------===//
|
|
//
|
|
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
|
|
// See https://llvm.org/LICENSE.txt for license information.
|
|
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#include "flang/Runtime/CUDA/descriptor.h"
|
|
#include "../terminator.h"
|
|
#include "flang/Runtime/CUDA/allocator.h"
|
|
#include "flang/Runtime/CUDA/common.h"
|
|
|
|
#include "cuda_runtime.h"
|
|
|
|
namespace Fortran::runtime::cuda {
|
|
extern "C" {
|
|
RT_EXT_API_GROUP_BEGIN
|
|
|
|
Descriptor *RTDEF(CUFAllocDesciptor)(
|
|
std::size_t sizeInBytes, const char *sourceFile, int sourceLine) {
|
|
return reinterpret_cast<Descriptor *>(
|
|
CUFAllocManaged(sizeInBytes, kCudaNoStream));
|
|
}
|
|
|
|
void RTDEF(CUFFreeDesciptor)(
|
|
Descriptor *desc, const char *sourceFile, int sourceLine) {
|
|
CUFFreeManaged(reinterpret_cast<void *>(desc));
|
|
}
|
|
|
|
void *RTDEF(CUFGetDeviceAddress)(
|
|
void *hostPtr, const char *sourceFile, int sourceLine) {
|
|
Terminator terminator{sourceFile, sourceLine};
|
|
void *p;
|
|
CUDA_REPORT_IF_ERROR(cudaGetSymbolAddress((void **)&p, hostPtr));
|
|
if (!p) {
|
|
terminator.Crash("Could not retrieve symbol's address");
|
|
}
|
|
return p;
|
|
}
|
|
|
|
void RTDEF(CUFDescriptorSync)(Descriptor *dst, const Descriptor *src,
|
|
const char *sourceFile, int sourceLine) {
|
|
std::size_t count{src->SizeInBytes()};
|
|
CUDA_REPORT_IF_ERROR(cudaMemcpy(
|
|
(void *)dst, (const void *)src, count, cudaMemcpyHostToDevice));
|
|
}
|
|
|
|
RT_EXT_API_GROUP_END
|
|
}
|
|
} // namespace Fortran::runtime::cuda
|