The PoCL open-source OpenCL implementation from Tampere University now has a new PoCL-Remote backend that enables transparent offloading of OpenCL tasks across nodes on a network, enabling distributing computing without using MPI or similar APIs. Since PoCL-Remote uses standard OpenCL API calls, acceleration offload can be performed identically whether using local or remote devices, for seamless selective/adaptive edge offloading and other use cases.
New PoCL-Remote Backend Enables Distributed Computing with Pure OpenCL - No MPI Needed
New PoCL-Remote Backend Enables Distributed Computing with Pure OpenCL - No MPI Needed