Gary Quinn (Gary++at++tees.ac.uk)
Thu, 29 Apr 1999 14:05:50 +0000
For info, I have (almost) a solution to the performance
problems under 6.5.3. Part one is a complete kludge,
as I wasn't able to trace the source of the problem in
the time I had available.
1. npri -r 128 -s RR command
This fixed the problem with some 3rd party s/w we're
using, which means that it was a CPU scheduling
problem, and not a gfx problem as I first thought.
More time spent with par might identify the exact
culprit, but I have to move on.
2. With our own apps,
pfuLockDownDraw()
pfuLockDownCull()
in the source code as well as npri has almost
restored the original performance. I guess
pfuLockDownApp() will complete the trick.
The frustrating thing is that these calls were
removed in February last year because they caused
some machines to hang completely :-(
I'd like to thank everybody who has provided suggestions,
in no particular order....
M.Y., Paul Hussein, Angus Dorbie
Cheers,
Gary.
--Gary Quinn, VR Centre Technical Manager University of Teesside SG Network Admin Borough Road, Middlesbrough Go Player Cleveland, TS1 3BA mailto:Gary++at++tees.ac.uk +44 (0)1642 384303 fax 384310 http://vr.tees.ac.uk
This archive was generated by hypermail 2.0b2 on Thu Apr 29 1999 - 06:06:01 PDT