AGP 8x slower than 4x on Radeon 9200SE?

"Alex Jackson" <[email protected]>
Newsgroups gmane.comp.video.dri.user
Message-ID <[email protected]>
I've just noticed that setting my AGP transfer speed (via BIOS setup) to 4x 
rather than 8x results in *faster* framerates in OpenGL apps with my Radeon 
9200SE.  The speed difference is a nearly constant 20% across the board, 
regardless of what OpenGL functionality the app uses:

glxgears:
AGP 4x: ~1900 FPS
AGP 8x: ~1500 FPS

sdlmame (unthrottled, running pacman):
AGP 4x: ~1100 FPS
AGP 8x: ~900 FPS

Note that these two apps are pretty much at two opposite ends of OpenGL 
usage: glxgears is all geometry and no textures, while sdlmame is doing 
nothing but continuously uploading a large texture and rendering it parallel 
to the screen.

I've looked in /var/log/messages and /var/log/Xorg.0.log and it looks like 
both AGP modes really are using the mode I select; neither is failing and 
falling back to a slower mode.  I'm running stock FC5 with all updates.

Does anyone have any idea why AGP 8x yields worse performance than 4x?  Does 
this happen to other people with rv280-based cards (the only chip in the 
r200 family that supports AGP 8x)?

--AWJ--

_________________________________________________________________
Off to school, going on a trip, or moving? Windows Live (MSN) Messenger lets 
you stay in touch with friends and family wherever you go. Click here to 
find out how to sign up!  http://www.live.com/?mkt=en-ca


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
--
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.