PGI and HPL benchmarking
Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.
Daniel Kidger Daniel.Kidger at quadrics.comWed Sep 19 11:06:28 PDT 2001
- Previous message: ALINKA Linux Clustering Letter, September 19th. 2001 (fwd)
- Next message: query about mpi neural network computing
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
I wrote: >You do not say why you want to use PGI's BLAS ? > >Check out >http://computational-battery.org/Programvare/blas-lib-comparison.html > >Their results show that Portland's BLAS implementation trails a long way >behind ATLAS and Intel's MKL. I investiagted a little further. It seems that PGI's BLAS is compiled up with backward compatability so that it will run on Pentium II's and lesser. Hence it makes no use of PIII and P4 instructions. i.e. no SSE / SSE2 or prefetching. There is a note in the release notes about recompiing their BLAS, but my distribution doesn't seem to have their source. Daniel. -------------------------------------------------------------- Dr. Dan Kidger, Quadrics Ltd. daniel.kidger at quadrics.com One Bridewell St., Bristol, BS1 2AA, UK 0117 907 5375 ----------------------- www.quadrics.com --------------------
- Previous message: ALINKA Linux Clustering Letter, September 19th. 2001 (fwd)
- Next message: query about mpi neural network computing
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
