RmmCudaAsyncMemoryResource#

public class RmmCudaAsyncMemoryResource implements RmmDeviceMemoryResource#

A device memory resource that uses cudaMallocAsync and cudaFreeAsync for allocation and deallocation.

Constructor Summary

Method Summary

close()

getHandle()

Returns a pointer to the underlying C++ class that implements an RMM device memory resource.

getSize()

toString()

Constructors

public RmmCudaAsyncMemoryResource(long size, long releaseThreshold)#

Create a new async memory resource.

Parameters:
  • size – the initial size of the pool

  • releaseThreshold – size in bytes for when memory is released back to cuda

public RmmCudaAsyncMemoryResource(
long size,
long releaseThreshold,
boolean fabric
)#

Create a new async memory resource.

Parameters:
  • size – the initial size of the pool

  • releaseThreshold – size in bytes for when memory is released back to cuda

  • fabric – if true request peer read+write accessible fabric handles when creating the pool

Methods

public long getHandle()#

Returns a pointer to the underlying C++ class that implements an RMM device memory resource.

public long getSize()#
public void close()#
public String toString()#