Memory resource that uses cudaMallocFromPoolAsync/cudaFreeFromPoolAsync with a managed memory pool for allocation/deallocation. More...

#include <cuda_async_managed_memory_resource.hpp>

Inheritance diagram for rmm::mr::cuda_async_managed_memory_resource:

Public Member Functions
	cuda_async_managed_memory_resource ()
	Constructs a cuda_async_managed_memory_resource with the default managed memory pool for the current device. More...

cudaMemPool_t	pool_handle () const noexcept
	Returns the underlying native handle to the CUDA pool. More...

	cuda_async_managed_memory_resource (cuda_async_managed_memory_resource const &)=default
	Default copy constructor.

	cuda_async_managed_memory_resource (cuda_async_managed_memory_resource &&)=default
	Default move constructor.

cuda_async_managed_memory_resource &	operator= (cuda_async_managed_memory_resource const &)=default
	Default copy assignment operator. More...

cuda_async_managed_memory_resource &	operator= (cuda_async_managed_memory_resource &&)=default
	Default move assignment operator. More...

Friends
void	get_property (cuda_async_managed_memory_resource const &, cuda::mr::device_accessible) noexcept
	Enables the `cuda::mr::device_accessible` property.

void	get_property (cuda_async_managed_memory_resource const &, cuda::mr::host_accessible) noexcept
	Enables the `cuda::mr::host_accessible` property.

Detailed Description

Memory resource that uses cudaMallocFromPoolAsync/cudaFreeFromPoolAsync with a managed memory pool for allocation/deallocation.

Constructor & Destructor Documentation

◆ cuda_async_managed_memory_resource()

rmm::mr::cuda_async_managed_memory_resource::cuda_async_managed_memory_resource ( )

Constructs a cuda_async_managed_memory_resource with the default managed memory pool for the current device.

The default managed memory pool is the pool that is created when the device is created. Its release threshold is set to the maximum value of std::uint64_t. This change is visible to all users of the device's default managed pool and retains unused backing memory across synchronizations.

Exceptions

rmm::logic_error if the CUDA version does not support cudaMallocFromPoolAsync with managed memory pool

Member Function Documentation

◆ operator=() [1/2]

cuda_async_managed_memory_resource& rmm::mr::cuda_async_managed_memory_resource::operator= ( cuda_async_managed_memory_resource && )

default

Default move assignment operator.

Returns: cuda_async_managed_memory_resource& Reference to the assigned object

◆ operator=() [2/2]

cuda_async_managed_memory_resource& rmm::mr::cuda_async_managed_memory_resource::operator= ( cuda_async_managed_memory_resource const & )

default

Default copy assignment operator.

Returns: cuda_async_managed_memory_resource& Reference to the assigned object

◆ pool_handle()

cudaMemPool_t rmm::mr::cuda_async_managed_memory_resource::pool_handle ( ) const

noexcept

Returns the underlying native handle to the CUDA pool.

Returns: cudaMemPool_t Handle to the underlying CUDA pool

The documentation for this class was generated from the following file:

cuda_async_managed_memory_resource.hpp

Public Member Functions