In a heterogeneous compute environment, developers must understand the capabilities and limitations of each architecture to effectively match the appropriate workload to each compute device.
In this session, oneAPI technical evangelist Karl Qi unpacks the unique and distinct programming needs of CPUs, GPUs, and FPGAs, including:
- Comparing and contrasting the architectural differences between CPUs, GPUs, and FPGAs
- Showing how Data Parallel C++ (DPC++) language constructs are mapped to each architecture
- Examining the support difference among oneAPI libraries—oneDPL, oneMKL, oneTBB, oneDAL, and more
- Discussing characteristics of applications best suited for each architecture