TensorRT-LLMs/cpp/tensorrt_llm/thop/torchAllocator.cpp
2023-10-10 23:22:17 -07:00

75 lines
2.2 KiB
C++

/*
* Copyright (c) 2022-2023, NVIDIA CORPORATION. All rights reserved.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "torchAllocator.h"
#include "tensorrt_llm/common/assert.h"
#include "tensorrt_llm/common/cudaUtils.h"
#include "tensorrt_llm/thop/thUtils.h"
using namespace tensorrt_llm::thop;
using namespace tensorrt_llm::common;
ReallocType TorchAllocator::reallocType(void const* ptr, size_t size) const
{
TLLM_CHECK(contains(ptr));
size_t currentSize = 1;
at::Tensor const& tensor{mPointerMapping.at(ptr)};
for (int i = 0; i < tensor.dim(); i++)
{
currentSize *= tensor.size(i);
}
TLLM_LOG_DEBUG("current_buffer_size: %d, original buffer: %p, new buffer: %d", currentSize, ptr, size);
if (currentSize < size)
{
return ReallocType::INCREASE;
}
else if (currentSize == size)
{
return ReallocType::REUSE;
}
else
{
return ReallocType::DECREASE;
}
}
void* TorchAllocator::malloc(size_t size, bool const setZero)
{
TLLM_LOG_DEBUG(__PRETTY_FUNCTION__);
auto const bufSize = static_cast<int64_t>(size);
torch::Tensor buf = torch::empty({bufSize}, torch::dtype(torch::kUInt8).device(torch::kCUDA));
void* ptr{buf.data_ptr()};
if (setZero)
{
memSet(ptr, 0, size);
}
TLLM_LOG_DEBUG("malloc buffer %p with size %ld", ptr, size);
mPointerMapping.insert({ptr, buf});
return ptr;
}
void TorchAllocator::free(void** ptr)
{
TLLM_LOG_DEBUG(__PRETTY_FUNCTION__);
mPointerMapping.erase(*ptr);
*ptr = nullptr;
}
void TorchAllocator::memSet(void* ptr, int const val, size_t const size)
{
check_cuda_error(cudaMemsetAsync(ptr, val, size, mStream));
}