MemPool#

classtorch.cuda.memory.MemPool(*args,**kwargs)[source]#

MemPool represents a pool of memory in a caching allocator. Currently,it’s just the ID of the pool object maintained in the CUDACachingAllocator.

Parameters

• allocator (torch._C._cuda_CUDAAllocator,optional) – atorch._C._cuda_CUDAAllocator object that can be used todefine how memory gets allocated in the pool. IfallocatorisNone (default), memory allocation follows the default/current configuration of the CUDACachingAllocator.

• use_on_oom (bool) – a bool that indicates if this pool can be usedas a last resort if a memory allocation outside of the pool fails dueto Out Of Memory. This is False by default.

propertyallocator:Optional[_cuda_CUDAAllocator]#

Returns the allocator this MemPool routes allocations to.

propertyid:tuple[int,int]#

Returns the ID of this pool as a tuple of two ints.

snapshot()[source]#

Return a snapshot of the CUDA memory allocator pool state across alldevices.

Interpreting the output of this function requires familiarity with thememory allocator internals.

Note

SeeMemory management for more details about GPU memorymanagement.

use_count()[source]#

Returns the reference count of this pool.

Return type

int