Combine Cuda IPC wth Direct Access

[eedalong]

I have 2 GPUs and they are under the same PCIe Switch. When I enable P2P access and allocate memory blocks on device1 and device0, I can access device1's memory block in kernels launched on device 0 within current process. I wander whether a cuda IPC memory handle, allocated on device1 from another process, can somehow be directly accessed in kernel launched on device0 within current process.

[inducer]

Isn't that what's controlled by the LAZY_ENABLE_PEER_ACCESS flag passed to the IPCMemoryHandle constructor?

See

• https://documen.tician.de/pycuda/driver.html?highlight=ipc#pycuda.driver.IPCMemoryHandle
• https://documen.tician.de/pycuda/driver.html?highlight=ipc#pycuda.driver.ipc_mem_flags.LAZY_ENABLE_PEER_ACCESS

[eedalong]

Yes, It works, Thx! @inducer