[comp.parallel] Benchmarking for SIMD/MIMD Archiectures

eugene@eos.arc.nasa.gov (Eugene Miya) (12/21/89)

The collection of benchmarks in the Perfect club are NOT tuned for
parallel archiectures.  They are just uniprocessor benchmarks.  So
too are the nbslib and netlib (ones that I have see, unless perhaps
you have a Fortran to dataflow compiler 8).


Marshall summarized the problems in his posting.  So this must of been
what it was like to program in the early days of the ENIAC? 8)
Well, less seriously so.  Wish I could help you, but there seems to
be a distinct lack of communications between the different
communities.  We crawl before we walk.

--enm