Re: CCL:re: Perf. of Alpha vs. R10k



 On Sep 23,  3:43pm, Chuck (aka chazix), 508-493-1399 (DTN 223-1399) wrote:
 > Subject: CCL:re: Perf. of Alpha vs. R10k
 >
 > I don't follow this forum, but a colleague forwarded Dr. van der Spoel's
 > notes.  Was the Fortran code that was posted (routine FORLJC, with a test
 > harness) the same code that produced a timing of 31 seconds on a 400MHz
 > Alpha?
 > On my system, it runs in less than 4 seconds, either with the options
 > Dr. van der Spoel gave, or with just "f90 -fast -tune host
 -non_shared".
    Hi,
     It runs in 1.3 seconds on an Indigo2 R10K machine (1 MB cache) with -Ofast,
 or 1.8 seconds using the compile options listed.  His 30 second time was
 presumably for the whole program.  It sounds like the R10K SGI machines are
 still significantly faster than the alpha machines even with the test code.
     The R10K chips were a BIG jump in performance over the previous R4xxx
 series
 from SGI, so this isn't that far from possibility.
     Take care,
        Dan
 --
 Dr. Daniel L. Severance     dan -8 at 8- sage.syntex.com       *****************
 Research Scientist          Work phone:(650) 354-7509  Note the Change
 Roche Bioscience (Syntex)   Fax (Work):(650) 852-1875  in area code!
 R6W-201, 3401 Hillview Ave  Palo Alto, CA  94304      *****************