HI All,
We're getting horrible thruput on some simple stuff, with GCG/WPI
running on a brand new Convex 3480. What I'd like to do is try some
equivalent stuff on an SGI to see if it's the CPU (doubt it) or the OS and
queueing problems (which I strongly suspect). The Convex has a very
sophisticated scheduler that makes it very efficient for using every cycle
but which makes it relatively unresponsive for interactive work, so even
simple jobs take relatively long times to get to completed. For instance
a restriction digest of lambda takes about 4 s on my Mac, and about 10
times as long on the convex in terms of wall clock time.
So, does anyone out there in GCG land have GCG/WPI running on an SGI of
just about any stripe that I might be able to borrow a few cycles of late
at night to evaluate whether the SGI is a better platform to host GCG for
the campus. Geographically closer to the LA area is better for the X
transport, but beggars....
Thanks in advance
Harry
--
Harry J Mangalam, Microbiology and Molecular Genetics, UC Irvine,
Irvine, CA, 92717, (714) 824-4824, fax (714) 824 8598
--- knowledge is fractal ---
http://hornet.mmg.uci.edu/~hjm/hjm.html
Computational Biology..SGI..Woodworking..Bicycling..Linux..WWW
