High-performance computing with desktop workstations
MATH'06 Proceedings of the 10th WSEAS International Conference on APPLIED MATHEMATICS
The "Chimera": an off-the-shelf CPU/GPGPU/FPGA hybrid computing platform
International Journal of Reconfigurable Computing - Special issue on High-Performance Reconfigurable Computing
| Hi-index | 0.00 |
This paper presents a new method to detect gamma-raypulsars using a fast folding algorithm [12] mapped ontoreconfigurable hardware. In contrast, existing techniquesrequire gigapoint complex FFTs. The algorithm has beenwritten in Streams-C and compiled with the sc2 compiler tothe target Annapolis Micro Systems (AMS) Firebird board(Xilinx Virtex E processor). To accelerate detection of newgamma-ray pulsars, the sc2 compiler generates a hardwareimplementation of the algorithm for finding periodicities indata sets. The data to be analyzed comes from a high energygamma-ray telescope onboard a spacecraft. This astro-physicsapplication poses a "good example" of the use of ahigh level reconfigurable computing tool such as sc2 to acceleratean algorithm because it uses real satellite data, thealgorithm can be parallelized, and was originally validatedusing a high level scientific language, IDL. By recasting thealgorithm into Streams-C, the scientific software developercan create a hardware implementation on a reconfigurablecomputing platform. We describe the fast folding algorithm,the Streams-C implementation, and discuss techniques tooptimize performance within the Streams-C framework. Thecompiler-generated hardware delivers approximately 3X to6X speed up over a comparable 800MHz general purposeprocessor doing the software-only algorithm.