Computer Architecture and Automated Design Lab

Using FPGAs for computational biochemistry and bioinformatics

As Field Programmable Gate Arrays (FPGAs) continue to get more powerful, they can be used in computational solutions to ever more complex problems. Even now they are approaching the point where grand-challenge-type applications can be addressed with FPGA-based computer systems: speed-ups of 1000x (over personal computers) have often been reported using a single FPGA. Use of clusters of PCs augmented with FPGAs is beginning to be investigated.

At the same time, the need for more cost-effective, flexible, and convenient high-performance computing is a common thread across diverse areas of Bioinfomatics and Computational Biology (BCB).

Our research has two aspects:

1. Proof-of-concept. We have examined and found BCB to be an extraordinarily target rich environment. We have found the following applications amenable to very large speed-ups using FPGA-based systems: microarray analysis, gene regulatory networks, modeling molecular interactions, molecular dynamics, various combinatorial optimization algorithms, as well as the already well-understood sequence processing algorithms.

2. Making the hardware transparent to application programmers. The key technology to be developed is the software environment This environment provides a three part solution. The first is a template layer, or set of high-level abstractions, to support the end user and the application programmer in focusing on the computational logic of the application with respect to the appropriate architectural model. The second is a module creation layer, built on top of a commercial hardware description language (VHDL), to enable efficient system construction (template assembly) based on methods of customizable logic components. The third is a set of automatic architectural optimizations built into the template layer, including replication, speed-matching, and precision management.

Accelerated Fault Tolerant Computing in Space with Reconfigurable Circuits

With funding from the U.S. Naval Research Laboratory, we are developing techniques for implementing fault-tolerace for FPGA acceleration of space-based applications. Satellites are exposed to radiation in space, and that can change the settings of the RAM cells that hold the FPGA's configuration. Our current work combines three module redundancy (TMR) with ability to reconfigure part of the FPGA while the other parts keep running. This allows us to update the FPGA configuration in each module in turn, while the other two modules continue the computation and error checking.

Interactions

• Sandor Vajda,
• Joe Mellor,
• Temple Smith (director of the BioMolecular Engineering Research Center BMERC),
• Charles Delisi and the Biomolecular Systems Lab (BSL),
• Zhiping Weng and her lab,
• Simon Kasif and the Computational Genomics Lab, and
• Gary Benson.