nV News Forums

 
 

nV News Forums (http://www.nvnews.net/vbulletin/index.php)
-   NVIDIA Linux (http://www.nvnews.net/vbulletin/forumdisplay.php?f=14)
-   -   Nvidia and SETI (yep) (http://www.nvnews.net/vbulletin/showthread.php?t=34352)

NotNvidious 08-09-04 10:46 AM

Nvidia and SETI (yep)
 
I am running SETI@Home on two Linux boxes. Box A is an Athlon XP 2300 with a GeForce FX 5200 card. Box B is a 2.4 GHz Pentium 4 with an Intel 82845G/GL graphics chipset. On A I have the 6106 Nvidia drivers, whereas on B I have Intel's i810 drivers, as delivered with the X Org X11 distribution. On box A glxgears yields some 1,800 fps, whereas on box B the performance is 1,300 fps (yee-ha!)

When I run the same tests with a SETI@Home client running on each box, things change dramatically. For A, glxgears reports 150 fps at best, whereas for B the number is about 750 fps. This raises tow questions:

1) How come that SETI@Home has such an enormous impact on the Nvidia drivers, and comparatively less so on the Intel ones?

2) Why, in either case, is the performance impact so large? In my naiveté, I thought that with the drivers above the graphics cards would be doing all the hard work, while leaving the CPU available for other chores. Obviously things are not quite as straightforward.

Anybody care to enlighten me here?

Thunderbird 08-09-04 11:14 AM

Re: Nvidia and SETI (yep)
 
Likely somehow the SETI client is quite active in the sense of taking a lot of cpu power. Check the amount of cpu power used by seti using things like top. Also check how much cpu it uses when both glxgears and SETI are running.

NotNvidious 08-09-04 12:16 PM

Re: Nvidia and SETI (yep)
 
Thanks for your reply. Yes, the SETI client is very active, and taking up all the CPU time when nothing else is running, and some %10-15% when glxgears is running - the rest is used by glxgears.

My original questions still remain though.

LordMorgul 08-09-04 05:49 PM

Re: Nvidia and SETI (yep)
 
I consider your results consistent with expectations. The Intel chip is substantially faster, and glxgears is effected alot by cpu load.. despite hardware support by the drivers that program is not being run only in hardware. Its purpose is only to verify that GL support is working.

Are these identical distributions, running kernels configured the same way? Are no cpu optimizations used? The intel is not running HT is it?

NotNvidious 08-09-04 10:54 PM

Re: Nvidia and SETI (yep)
 
Quote:

Originally Posted by LordMorgul
I consider your results consistent with expectations. The Intel chip is substantially faster,

The truth is that my Athlon box actually feels faster.

Quote:

Originally Posted by LordMorgul
and glxgears is effected alot by cpu load.. despite hardware support by the drivers that program is not being run only in hardware. Its purpose is only to verify that GL support is working.

I am just intrigued at the relative impact on performance when using the Athlon/Nvidia combination, when compared against the Pentium IV/Intel i845G one.

Quote:

Originally Posted by LordMorgul
Are these identical distributions, running kernels configured the same way?

They are both Slackware 10. The kernels are not exactly the same, due to hardware differences between platform, as you can imagine.


Quote:

Originally Posted by LordMorgul
Are no cpu optimizations used? The intel is not running HT is it?

No on both counts.


All times are GMT -5. The time now is 06:14 PM.

Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright ©1998 - 2014, nV News.