Merrimac: Supercomputing with Streams

William J. Dally, Patrick Hanrahan, Mattan Erez, Timothy J. Knight, François Labonté, Jung-Ho Ahn, Nuwan Jayasena, Ujval J. Kapasi, Abhishek Das, Jayanth Gummaraju, and Ian Buck

in the Proceedings of the SC'03 Conference, November 2003, Phoenix, Arizona.

Abstract:

Merrimac uses stream architecture and advanced interconnection networks to give an order of magnitude more performance per unit cost than cluster-based scientific computers built from the same technology. Organizing the computation into streams and exploiting the resulting locality using a register hierarchy enables a stream architecture to reduce the memory bandwidth required by representative applications by an order of magnitude or more. Hence a processing node with a fixed bandwidth (expensive) can support an order of magnitude more arithmetic units (inexpensive). This in turn allows a given level of performance to be achieved with fewer nodes (a 1-PFLOPS machine, for example, with just 8,192 nodes) resulting in greater reliability, and simpler system management. We sketch the design of Merrimac, a streaming scientific computer that can be scaled from a $20K 2 TFLOPS workstation to a $20M 2 PFLOPS supercomputer and present the results of some initial application experiments on this architecture.

Paper:

Adobe Acrobat PDF

BibTeX:

@CONFERENCE{ref:merrimac_sc03,
   author = {W. J. Dally and P. Hanrahan and M. Erez and T. J. Knight and F. Labonte and J-H A. and N. Jayasena and U. J. Kapasi and A. Das and J. Gummaraju and I. Buck},
   title = {Merrimac: Supercomputing with streams},
   booktitle = {{SC'}03},
   year = {2003},
   address = {Phoenix, Arizona},
   month = {November}
}