OpenCL

Optimizing AI workloads on heterogeneous hardware platforms is currently driving discussions.