InAccel Tutorial: How to scale your FPGA design in the cloud or on-prem, instantly

How to scale your FPGA design on the cloud or on-prem, instantly

Abstract

everal cloud providers like AWS, Alibaba and Huawei have started deploying FPGAs in their cloud. This is a great opportunity for the FPGA community that can deploy their FPGA designs on the cloud and through the relevant marketplaces to commercialize their design all over the world. However, the current framework provided by the cloud and the FPGA vendors have certain limitations that make it very complicated when you try to scale your FPGA in the cloud. In this tutorial we will show how you can port your FPGA design on the cloud using the tools available and then how you can scale-up and scale-out your design using the free Community edition of the InAccel FPGA resource manager. In this tutorial we will demonstrate how easy it is to scale-up your FPGA design from 1 FPGA (f1.2x) to multiple FPGAs (f1.4x or f1.16x) instantly and then how to scale out your FPGA design to multiple servers using framework like Apache Spark. Finally, we will show how to use your FPGA designs to address the current challenges in the domain of big data and machine learning applications in which high computational power is required.

Agenda

  • ♦ Porting an FPGA design on the cloud or on-prem

  • ♦ Scaling your FPGA design to multiple FPGAs on the same server (f1.2x to f1.16x)

  • ♦ Integrating your FPGA with high level frameworks and scaling your FPGA design to multiple servers (Kubernetes)

  • ♦ Sharing and virtualization of the FPGA cluster

Organizer - InAccel

InAccel is a leader in application acceleration using FPGAs in the cloud. InAccel helps companies speedup their applications in the cloud and reduce their TCO by providing ready-to-use high-performance accelerators. InAccel is one of the first companies that provided accelerators as a service in the cloud for FPGAs and is an official partner of Amazon AWS, Xilinx and Intel PSG.