warp.set_mempool_access_enabled

warp.set_mempool_access_enabled(target_device, peer_device, enable)

Enable or disable access from peer_device to the memory pool of target_device.

This applies to memory allocated using CUDA pooled allocators. For memory allocated using default CUDA allocators, use set_peer_access_enabled().

Parameters:

• target_device (Device | str | None)

• peer_device (Device | str | None)

• enable (bool)

Return type:

None