Power-Efficient Machine Learning on POWER Systems using FPGA Acceleration



This session will provide an overview of how FPGA acceleration can enhance POWER systems for machine learning workloads such as image recognition.  Both recent results from various field implementations and extrapolation of throughput per watt expected in the future will be discussed.  Additionally, a brief background of Xilinx FPGA technology as it applies to developers of machine learning systems will also be discussed, including discussion of CAPI interconnect used in POWER systems.

Speaker Bio

Ralph Wittig is Distinguished Engineer at Xilinx. He works on heterogeneous processing platforms combining FPGAs with CPUs, GPUs, and DSPs and he drives FPGA programming tools based on open standards such as OpenCL and OpenMP. Wittig has received an M.A.Sc. degree in Computer Engineering from the University of Toronto. He is the holder of 40+ patents.