56fd68e7dd
CCCL (CUDA C++ Core Libraries) provides: - CUB: device/block/warp-level GPU primitives (reduce, scan, sort, topk) - Thrust: high-level parallel algorithms (transform_reduce, sort, scan) - libcudacxx: CUDA C++ standard library (atomics, barriers, memory) - cudax: experimental features (memory resources, allocators) - Tuning policies: per-SM hardware-specific algorithm parameters Competition optimization vectors mapped to CCCL: - Output TPS (83% weight): warp_reduce, block_reduce, device_topk - Input TPS (14% weight): device_scan, block_load, prefetch - Cache TPS (3% weight): prefix caching strategy patterns - Memory (0.9 util): pooled/cached/buddy allocators Source: https://github.com/NVIDIA/cccl (shallow clone, HEAD only) License: Apache-2.0
50 lines
1.3 KiB
C++
50 lines
1.3 KiB
C++
// SPDX-FileCopyrightText: Copyright (c) 2018, NVIDIA Corporation. All rights reserved.
|
|
// SPDX-License-Identifier: Apache-2.0
|
|
|
|
/*! \file tls_pool.h
|
|
* \brief A function wrapping a thread local instance of a \p unsynchronized_pool_resource.
|
|
*/
|
|
|
|
#pragma once
|
|
|
|
#include <thrust/detail/config.h>
|
|
|
|
#if defined(_CCCL_IMPLICIT_SYSTEM_HEADER_GCC)
|
|
# pragma GCC system_header
|
|
#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_CLANG)
|
|
# pragma clang system_header
|
|
#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_MSVC)
|
|
# pragma system_header
|
|
#endif // no system header
|
|
#include <thrust/mr/pool.h>
|
|
|
|
THRUST_NAMESPACE_BEGIN
|
|
namespace mr
|
|
{
|
|
/*! \addtogroup memory_resources Memory Resources
|
|
* \ingroup memory_management
|
|
* \{
|
|
*/
|
|
|
|
/*! Potentially constructs, if not yet created, and then returns the address of a thread-local \p
|
|
* unsynchronized_pool_resource,
|
|
*
|
|
* \tparam Upstream the template argument to the pool template
|
|
* \param upstream the argument to the constructor, if invoked
|
|
*/
|
|
template <typename Upstream, typename Bookkeeper>
|
|
_CCCL_HOST thrust::mr::unsynchronized_pool_resource<Upstream>& tls_pool(Upstream* upstream = nullptr)
|
|
{
|
|
static thread_local auto adaptor = [&] {
|
|
assert(upstream);
|
|
return thrust::mr::unsynchronized_pool_resource<Upstream>(upstream);
|
|
}();
|
|
|
|
return adaptor;
|
|
}
|
|
|
|
/*! \}
|
|
*/
|
|
} // namespace mr
|
|
THRUST_NAMESPACE_END
|