Multi-Device called by multi cpu thread

[stabunkow]

Could ROCm consider non-lock hipMemcpyAsync API when each thread control a device?

[stabunkow]

#pragma omp for
for (deviceId = 0; deviceId < deviceSiz; ++deviceId) 
{
   hipSetDevice(deviceId);
   hipMemcpyAsync(...);
}

I know there is spinlock inside hipMemcpyAsync API, but when set hipSetDevice, the spinlock will be removed?

[ppanchad-amd]

@stabunkow Apologies for the lack of response. Do you still need assistance with this ticket? Thanks!