gpu_memcpy.hpp

Go to the documentation of this file.

/*
  Copyright 2026 Equinor ASA
 
  This file is part of the Open Porous Media project (OPM).
  OPM is free software: you can redistribute it and/or modify
  it under the terms of the GNU General Public License as published by
  the Free Software Foundation, either version 3 of the License, or
  (at your option) any later version.
 
  OPM is distributed in the hope that it will be useful,
  but WITHOUT ANY WARRANTY; without even the implied warranty of
  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  GNU General Public License for more details.
 
  You should have received a copy of the GNU General Public License
  along with OPM.  If not, see <http://www.gnu.org/licenses/>.
*/
 
#ifndef OPM_GPUISTL_DETAIL_GPU_MEMCPY_HPP
#define OPM_GPUISTL_DETAIL_GPU_MEMCPY_HPP
 
#include <opm/simulators/linalg/gpuistl/detail/gpu_pointer_attributes.hpp>
#include <opm/simulators/linalg/gpuistl/detail/gpu_safe_call.hpp>
#include <opm/simulators/linalg/gpuistl/detail/gpu_stream.hpp>
 
#include <cuda_runtime.h>
 
#include <cstddef>
 
namespace Opm::gpuistl::detail
{
 
template <typename T>
inline void
gpuMemcpyHostToDevice(T* dstDevice, const T* srcHost, std::size_t count)
{
    if (count == 0) {
        return;
    }
    OPM_GPUISTL_DETAIL_ASSERT_DEVICE_POINTER(dstDevice);
    OPM_GPUISTL_DETAIL_ASSERT_HOST_POINTER(srcHost);
    OPM_GPU_SAFE_CALL(cudaMemcpy(dstDevice, srcHost, count * sizeof(T), cudaMemcpyHostToDevice));
}
 
template <typename T>
inline void
gpuMemcpyDeviceToHost(T* dstHost, const T* srcDevice, std::size_t count)
{
    if (count == 0) {
        return;
    }
    OPM_GPUISTL_DETAIL_ASSERT_HOST_POINTER(dstHost);
    OPM_GPUISTL_DETAIL_ASSERT_DEVICE_POINTER(srcDevice);
    OPM_GPU_SAFE_CALL(cudaMemcpy(dstHost, srcDevice, count * sizeof(T), cudaMemcpyDeviceToHost));
}
 
template <typename T>
inline void
gpuMemcpyDeviceToDevice(T* dstDevice, const T* srcDevice, std::size_t count)
{
    if (count == 0) {
        return;
    }
    OPM_GPUISTL_DETAIL_ASSERT_DEVICE_POINTER(dstDevice);
    OPM_GPUISTL_DETAIL_ASSERT_DEVICE_POINTER(srcDevice);
    OPM_GPU_SAFE_CALL(
        cudaMemcpy(dstDevice, srcDevice, count * sizeof(T), cudaMemcpyDeviceToDevice));
}
 
template <typename T>
inline void
gpuMemcpyHostToDeviceAsync(T* dstDevice, const T* srcHost, std::size_t count, cudaStream_t stream)
{
    if (count == 0) {
        return;
    }
    OPM_GPUISTL_DETAIL_ASSERT_DEVICE_POINTER(dstDevice);
    OPM_GPUISTL_DETAIL_ASSERT_HOST_POINTER(srcHost);
    OPM_GPUISTL_DETAIL_ASSERT_GPU_STREAM(stream);
    OPM_GPU_SAFE_CALL(
        cudaMemcpyAsync(dstDevice, srcHost, count * sizeof(T), cudaMemcpyHostToDevice, stream));
}
 
template <typename T>
inline void
gpuMemcpyDeviceToHostAsync(T* dstHost, const T* srcDevice, std::size_t count, cudaStream_t stream)
{
    if (count == 0) {
        return;
    }
    OPM_GPUISTL_DETAIL_ASSERT_HOST_POINTER(dstHost);
    OPM_GPUISTL_DETAIL_ASSERT_DEVICE_POINTER(srcDevice);
    OPM_GPUISTL_DETAIL_ASSERT_GPU_STREAM(stream);
    OPM_GPU_SAFE_CALL(
        cudaMemcpyAsync(dstHost, srcDevice, count * sizeof(T), cudaMemcpyDeviceToHost, stream));
}
 
} // namespace Opm::gpuistl::detail
 
#endif // OPM_GPUISTL_DETAIL_GPU_MEMCPY_HPP