High performance computers from

• 4-rack Origin 2000 system
• 64 CPUs: 300 MHz R12000 processors
• 16 GBytes main memory (RAM)
• 1.3 TeraBytes disk space
• Hecate is shared among users in Departments of Astrophysics, Computer Sciences and Princeton Plasma Physics Lab
• Questions about hecate may be sent to bill@princeton.edu

Hecate Usage Guidelines: Read Before Running Jobs on Hecate

Performance Tuning for Origin 2000

• -64 -mips4 -r12000 -TARG:platform=ip27    optimization for the Origin 2000 architecture
• -Ofast=ip27    maximum optimization level for Origin 2000
• -OPT:IEEE_arithmetic=3 -OPT:roundoff=3    speeds up arithmetic when round-off errors are not important   (check that results are the same)
• -OPT:alias=disjoint
• -IPA=on    cross-subroutine optimization
• -LNO:vintr=on    automatic vectorization

• -lfastm    fast math functions

Parallel Programming Tutorials

• Users can use command "insight" on hecate to get online help
• "Introduction to Parallel Processing on SGI Shared Memory Computers", Boston University
• "Multiprocessing with Fortran 90 (and HPF)", Boston University
• "Introduction to Parallel Programming", University of Illinois UC
• "Parallel Numerical Algorithms", University of Illinois UC
• "Applications of Parallel Computers", UC Berkeley
• "Designing and Building Parallel Programs", by Ian Foster

Supercomputer Links

• National Center for Supercomputing Applications , Urbana-Champaign
• Pittsburgh Supercomputer Center
• Cornell Theory Center
• Maui High Performance Computing Center
• San Diego Supercomputing Center
• University of Michigan - Center for Parallel Computing
  
  
• List of parallel computing sites
  
  

• MPI: The Complete Reference
• MPICH - A portable MPI implementation
• in Designing and Building Parallel Programs, by Ian Foster
• Other resources from the Mississippi State University

• Fortran FLIB