FPGAs play a critical role in heterogeneous compute platforms as flexible, reprogrammable, multi-function accelerators. They enable custom-hardware performance with the programmability of software. The industry trend toward software-defined hardware challenges not just the traditional architectures—compute, memory, network resources—but also the programming model of heterogeneous compute platforms. Traditionally, the FPGA programming model has been narrowly tailored and hardware-centric. As FPGAs become part of heterogeneous compute platforms and users expect the hardware to be “software-defined,” FPGAs must be accessible not just by hardware developers but also by software developers, which requires the programming model of FPGAs to evolve dramatically.
This session details a highly evolved, software-centric programming model that enables software developers to harness FPGAs through a comprehensive solutions stack. It encompasses FPGA-optimized libraries, compilers, tools, frameworks, SDK integration and an FPGA-enabled ecosystem. Your training will also include real-world examples using machine learning inference acceleration on FPGAs.