Skip to main content
Home Forums 68kMLA AMD Athlon Tower — #8
Post #8 by QuadSix50
Source Forum68kMLA
CategoryHardware
Post DateThu, 3 Jul 2008 - 04:02
Original URLhttps://68kmla.org/bb/threads/amd-athlon-tower.13816/
Post
It's more of a "These, these, and these ATi cards are blacklisted and generally don't work well until we get better drivers" type thing. The card works fine, but there is no proper hardware rendering.
A new xorg.conf wont do a thing unfortunately.

EDIT: Also, that site is hilarious. They actually use glxgears as a serious benchmark tool. Hahaha.

But really, if I compile glxgears under some optimizations, I get 50fps, under others I get several thousand, with the same graphical output. Not a good tool to use for a benchmark.
Well, glxgears does show you whether you are getting decent hardware acceleration or not. I've used it to see how the card is rendering 3D and if it's being done by software instead of the hardware. According to the site, your card should do hardware rendering with the open source driver. Not sure if the same is the case with ATI's fglrx driver (I've never had much success with that). For the record, you shouldn't have to add anything in the "Section 'Device'" portion of xorg.conf other than to specify that you're using the "radeon" driver.

Code:
Graphics Card | CPU                     | X version   | Driver-Name | DRI? | GLX Server | Resolution | Depth | Notes | glxgears FPS 
===================================================================================================================================
ATI X300      | Intel Pentium 2.8GHz HT | X.Org 7.1.1 | radeon      | Yes  | v1.2       | 1280x1024  | 24    |       | 1550 FPS
Another thing I do is maximize the glxgear window. If it's being rendered in software, then the moment I maximize the window it would slow down tremendously. If I do the same thing and the hardware rendering is working, then it continues on at pretty much the same speed. Then again, you don't have to just use glxgears. There are other 3D based screensavers you can manually launch that are much more complex than glxgears. Sure, it's not a benchmarking tool, but it's a quick way to compare how quickly the hardware rendering is being done, if at all.

Also note the disclaimer on the page (which reaffirms what I said):

To gather the results in the table below, a simple shell script was proposed Henrik Carlqvist in a comp.os.linux.hardware discussion and subsequently revised by Michael Mauch and others. This script uses glxgears to get a frame rate for 3D rendering. We are aware this isn’t a particularly good benchmark. It’s been compared to bogomips for graphics cards. But it is readily available on almost every computer running Xorg and provides a handy baseline for comparing hardware. If you can propose something better that will run on as many systems without needing to download and compile special benchmarking tools, please let us know. But see the FAQ before emailing to complain that you think glxgears sucks as a benchmark.
mp.ls