cuda::experimental::pinned_memory_pool#

class pinned_memory_pool : public cuda::experimental::__memory_pool_base#

pinned_memory_pool is an owning wrapper around a cudaMemPool_t.

It handles creation and destruction of the underlying pool utilizing the provided memory_pool_properties.

Public Functions

inline explicit pinned_memory_pool(
int __numa_id,
memory_pool_properties __properties,
)#

Constructs a pinned_memory_pool with the specified NUMA node id and initial pool size and release threshold.

If the pool size grows beyond the release threshold, unused memory held by the pool will be released at the next synchronization event.

Note

Memory from this pool is accessible from all devices right away, which differs from the default behavior of pinned memory pools where memory is not accessible from devices until cudaMemPoolSetAccess is called.

Parameters:

  • __numa_id – The NUMA node id of the NUMA node the pool is constructed on.

  • __pool_properties – Optional, additional properties of the pool to be created.

inline explicit pinned_memory_pool(int __numa_id = 0)#

Constructs a pinned_memory_pool on the host with the specified NUMA node id.

Note

Memory from this pool is accessible from all devices right away, which differs from the default behavior of pinned memory pools where memory is not accessible from devices until cudaMemPoolSetAccess is called.

Parameters:

  • __numa_id – The NUMA node id of the NUMA node the pool is constructed on.

  • __numa_id – The NUMA node id of the NUMA node the pool is constructed on.

pinned_memory_pool(::cudaMemPool_t) = delete#

Disables construction from a plain cudaMemPool_t. We want to ensure clean ownership semantics.

pinned_memory_pool(pinned_memory_pool const&) = delete#
pinned_memory_pool(pinned_memory_pool&&) = delete#
pinned_memory_pool &operator=(pinned_memory_pool const&) = delete#
pinned_memory_pool &operator=(pinned_memory_pool&&) = delete#

Public Static Functions

static inline pinned_memory_pool from_native_handle(
::cudaMemPool_t __handle,
) noexcept#

Construct an pinned_memory_pool object from a native cudaMemPool_t handle.

Note

The constructed pinned_memory_pool object takes ownership of the native handle.

Parameters:

__handle – The native handle

Returns:

The constructed pinned_memory_pool object

static pinned_memory_pool from_native_handle(int) = delete#
static pinned_memory_pool from_native_handle(
::cuda::std::nullptr_t,
) = delete#