Abstract
Computational Fluid Dynamics (CFD) applications are a critical tool in designing sophisticated mechanical systems such as jet engines and gas turbines. CFD applications use intensive floating-point calculations and are typically run on High-Performance Computing (HPC) systems. We analyze three of the most compute intensive functions (Euler, Viscous, and Smoothing algorithms) and develop a baseline system architecture for accelerating these functions in RCC hardware. We then present detailed design data for the most compute intensive (Euler) function. Based on this analysis, we show that an RCC-based CFD accelerator—compared to conventional processors—promises dramatic improvement in sustained compute speed at better price-performance ratios coupled with much lower overall power consumption.
Similar content being viewed by others
References
Advantech ACP-7000 7U 20-slot rack mount chassis, http://www.advantech.com/
Advantech PCA-6119P16X 19-slot 64-bit Backplane; 2 ISA/16 PCI/1 CPU, 1 Segment, http://www. advantech.com/
Advantech PCA-6278 Dual Socket 370 Pentium III (Tualatin) Single Board Computer Card with 64-bit PCI/VGA/Dual LAN/HISA (133 MHz FSB), http://www.advantech.com/
Celoxica DK Design Suite, http://www.celoxica.com/
J. H. Ferziger and M. Peric. Computational Methods for Fluid Dynamics. Springer Verlag, New York, 2001.
J. Hammes. Methodology for pipelining and fusing stenciled loops, Technical paper SWP-009-00, SRC computers, Inc., Colorado Springs, Colorado, 2003.
D. Holmgren, A. Singh, P. Mackenzie, and J. Simone. Lattice QCD production on commodity clusters at fermilab. In Computing in High Energy and Nuclear Physics (CHEP03). SLAC and UCSD, La Jolla, California, March 24-28, 2003, http://www.slac.stanford.edu/econf/C0303241/
J. Makino. Massively parallel computer for particle-based simulations. 5th JSPS Symposium on Computational Science and Engineering, Japan Society for the Promotion of Science (JSPS), University of Tokyo, 2002. http://olab.is.s.u-tokyo.ac.jp/cse/index-e.html
J. Makino, E. Kokubo, et al. Performance evaluation and tuning of GRAPE-6-towards 40 "real" Tflops. In Proceedings of the ACM/IEEE SC2003 Conference, Phoenix, Arizona, 2003. CDROM and http://www.sc-conference. org/sc2003/
Nallatech Ben DATA-WS TM DIME-II TM module, http://www.nallatech.com/products/dime2/bendata-ws/ index.asp.
Nallatech Ben NUEY TM carrier card, http://www.nallatech.com/products/dime2/bennuey/index.asp
Nallatech FUSE TM environment, http://www.nallatech.com/products/fuse/index.asp
Nallatech web site, http://www.nallatech.com/
OctigaBay Systems Corporation's OctigaBay 12K processor, http://www.octigabay.com/productsServices/ product.htm
W. D. Smith and A. R. Schnore. Towards an RCC-based accelerator for computational fluid dynamics applications. In T. P. Plaks, ed., Proceedings of the International Conference of Engineering of Reconfigurable Systems and Algorithms, CSREA Press: Las Vegas, Nevada, pp. 222–231, 2003.
SRC Computers, Inc.'s MAP Processor, http://www.srccomp.com/MAPProc.htm
Xilinx FORGE, http://www.xilinx.com/ise/advanced/forge.htm
Xilinx ISE 5.1 Foundation tools, http://www.xilinx.com/
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Smith, W.D., Schnore, A.R. Towards an RCC-Based Accelerator for Computational Fluid Dynamics Applications. The Journal of Supercomputing 30, 239–261 (2004). https://doi.org/10.1023/B:SUPE.0000045211.07895.cb
Issue Date:
DOI: https://doi.org/10.1023/B:SUPE.0000045211.07895.cb