The Era of FPGA Acceleration

Recent advances in FPGAs, multi-core processors, and memory technologies have enabled high-performance platforms for accelerating complex applications. This talk reviews three decades of progress in reconfigurable computing and highlights current FPGA-based accelerators for data science. We present algorithm-architecture co-design techniques for developing efficient IP cores and end-to-end acceleration, with examples in graph analytics, machine learning, and privacy-preserving computations. Finally, we discuss opportunities and challenges in emerging heterogeneous systems combining CPUs, FPGAs, GPUs, and coherent memory.

This session is part of our “Innovators Day Webinar Series”. Registration/event page here.

Sponsored by: