Design and implementation of message-passing services for the Blue Gene/L supercomputer

George Almási, Charles Archer, José Gabriel Castaños, John A. Gunnels, C. Chris Erway, Philip Heidelberger, Xavier Martorell, José E. Moreira, Kurt Pinnow, Joseph Ratterman, Burkhard D. Steinmacher-Burow, William Gropp, Brian Toonen

Research output: Contribution to journalArticlepeer-review

Abstract

The Blue Gene®/L (BG/L) supercomputer, with 65,536 dual-processor compute nodes, was designed from the ground up to support efficient execution of massively parallel message-passing programs. Part of this support is an optimized implementation of the Message Passing Interface (MPI), which leverages the hardware features of BG/L. MPI for BG/L is implemented on top of a more basic message-passing infrastructure called the message layer. This message layer can be used both to implement other higher-level libraries and directly by applications. MPI and the message layer are used in the two BG/L modes of operation: the coprocessor mode and the virtual node mode. Performance measurements show that our message-passing services deliver performance close to the hardware limits of the machine. They also show that dedicating one of the processors of a node to communication functions (coprocessor mode) greatly improves the message-passing bandwidth, whereas running two processes per compute node (virtual node mode) can have a positive impact on application performance.

Original languageEnglish (US)
Pages (from-to)393-406
Number of pages14
JournalIBM Journal of Research and Development
Volume49
Issue number2-3
StatePublished - Jan 1 2005
Externally publishedYes

ASJC Scopus subject areas

  • Computer Science(all)

Fingerprint Dive into the research topics of 'Design and implementation of message-passing services for the Blue Gene/L supercomputer'. Together they form a unique fingerprint.

Cite this