62. RIFFA: A REUSABLE INTEGRATION FRAMEWORK FOR FPGA ACCELERATORS

Department: Computer Science & Engineering
Faculty Advisor(s): Ryan Kastner

Primary Student
Name: Matthew D Jacobsen
Email: mdjacobs@ucsd.edu
Phone: 510-410-9898
Grad Year: 2013

Abstract
We present RIFFA, a reusable systems level integration framework for FPGA accelerators. RIFFA provides communication and synchronization for FPGA accelerated software using a standard interface. Our goal is to expand the use of FPGAs as an acceleration platform by releasing, as open source, a no cost framework that easily integrates software on traditional CPUs with FPGA based IP cores, over PCIe, with minimal custom configuration. RIFFA requires no specialized hardware or fee licensed IP cores. It can be deployed on common Linux workstations with a PCIe bus and has been tested on two different Linux distributions using Xilinx FPGAs. We demonstrate RIFFA's usefulness through two use case video applications: a multi-point particle filter tracker and a Viola and Jones face detector. The tracker shows a 30x performance increase over the software only version. Similarly, the face detector frame rate is increased by 3.16x.

Related Links:

  1. http://seed.ucsd.edu/~mjacobsen/fpga_tracker.m4v
  2. http://seed.ucsd.edu/~mjacobsen/fpga_fd.m4v

« Back to Posters or Search Results