In article <950213163348.2040c062 at JAGUAR.CSC.WSU.EDU> THOMPSON at JAGUAR.CSC.WSU.EDU ("Steve Thompson: VADMS genetics") writes:
>>The College of Sciences and the Biochemistry/Biophysics Department at
>Washington State University are purchasing a new computer system to serve VADMS
>subscribers' computational needs. The system will be a Silicon Graphics Power
>Challenge L with the new R8000 chip and IRIX system 6.0.1. GCG has warned us
>that they do not support this new configuration. Do you have any machines
>similarly configured and have you been able to run GCG v.8.0 and/or the
>available EGCG v.8.0 programs on them? If it is possible, what type of
>problems have you encountered? Thanks for any advice that you can offer.
In theory, the Irix 5.x executables will run on a power challenge, and under
Irix 6.0.x, but you will not get particularly good performance with them.
Having said that, let me add that GCG is not particularly well suited for
running on the R8000 architecture. That chip was designed with Floating
Point performance in mind, and programs like GCG which (to the best of my
knowledge) are primarily integer based, will not speed up as much. You would
see some small improvement from the increased cache size, but nothing like the
benchmark numbers SGI provides (which are floating point benchmarks usually).
The bottom line, I guess, is that if you are going to be doing structural
refinements and other FP intensive number-crunching, the Power Challenge is
a great platform (We have one here ourselves). GCG may not officially support
the OS, but as long as they aren't doing anything TOO perverse with their
source, it should run. If you are only going to be doing sequence analysis, a
Power Challenge might be overkill for your needs and you might be happier with
a regular Challenge server.
Rice University Biochemistry and Cell Biology
fant at bioc.rice.edu
(.sig still under renovation)