cuda::experimental::stf::uncached_block_allocator
Defined in include/cuda/experimental/__stf/allocators/uncached_allocator.cuh
-
class uncached_block_allocator : public cuda::experimental::stf::block_allocator_interface
Uncached block allocator where allocations and deallocations are directly performed as CUDA API calls (cudaMallocAsync, …)
This is actually a wrapper on top of the uncached allocator automatically created within each context backend, so deinit is a no-op. The intent of this wrapper is to make it possible to create an allocator policy using this internal allocator.
Public Functions
-
inline virtual void print_info() const
Print information about the allocator.
This method prints information about the allocator to stderr.
-
inline virtual void print_info() const