Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×

Journal gnarlin's Journal: Why do Free software driver perform so badly? 3

I recently installed Ubuntu feisty fawn (7.04) on my laptop.
It's a Mitac 8050D, has 1GB ram, a 1.5Ghz pentium M, and a Radeon9700 mobility and a 60GB hard drive. I like it. Everything works right out of the box. Altough the ipw2200 does have to load a non-free blob firmware for the wireless to work.

The problem is that I like playing games and no, I don't want to boot Microsoft Windows. There is a myth that there aren't any fun or decent games, let alone Free games for the gnu/linux platform. The truth is that although they aren't too numerous there are several which I enjoy quite a bit. In order to be able to play advanced 3d first person shooters like Nexuiz or Warsow with enough FPS to be competetive I have to load the fglrx propriatery driver. There is a Free driver for my radeon9700 but it is just isn't good enough performance wise. I read an article recently at phoronix.com in which the Freedom giving drivers were compaired against the propriatery drivers from ATI. The closed one gave the Free one a sound beating. This article follows up on an older article in which they did the exact same thing. The conclusion is that there has been negliable increase in performance of the Free drivers since the last time. What I want to know is why there seems to be close to no change in those drivers performance? Have the effords to reverse engineer the r300 chips been stopped because the radeon driver is deemed "good enough" for basic desktop needs?

For a long time people have hoped that nvidia would throw the community a bone by publishing at least some specs for the graphic chips. No such luck. At long last the community got fed up with waiting. There is now an effort afoot to reverse engineer their cards and to write some Free drivers and progress has been going quite well. The problem that I see is that once those drivers work "well enough" for basic desktop usage, namely that they can run compiz/beryl decently, that development will slow down to a crawl and that they will therefore never be competitive performance wise with the "official" nvidia drivers. I don't just mean that they will lag slighty behind, but more like I won't be able to play a single game in opengl with enough FPS to be tolerable.

At this point someone might get a little irrited that I am slinging mud at the people who are spending their free time to give to the community, so perhaps I should make my position clear. I like Free software. In fact I prefer it but I'll use non-Free software if I really have to (if there is no decent Free alternative). But the reality is that I really like playing games and that I am not rich enough to be able to just buy a new one with a graphic chip which has good enough Free drivers available so I do what I must to play the games. There is hope with Intel recently opening the drivers up to the community. Their chips might not be the best, but they might be good enough for games. The people developing the nouveau nvidia driver and the people at dri and mesa have my adoration and respect. They are doing the world a huge favor by working on this stuff. I hope that they succeed beyond everyone's wildest dreams and end up making the drivers betters than everything else. I just hope that it won't take forever.

This discussion has been archived. No new comments can be posted.

Why do Free software driver perform so badly?

Comments Filter:
  • Hardware chips are the cumulative products of thirty or forty years of software engineering. For business reasons, a company that has money invested in the software isn't going to give the specs out to just anyone. At the same time they are going to make it as difficult as possible to reverse engineer the functions of the hardware chips. Writing the firmware and drivers to make the hardware chips accessible, through standardized software libraries, to software writers (game developers) takes lots and lot
    • by gnarlin ( 696263 )
      Well, if it truly is the result of a 20-30 years of effort that have gone into these graphic cards and reverse engineering them is as difficult as you describe it then perhaps it is just easier to do what the opengraphics.org [opengraphics.org] people are doing. Starting from scratch and make new hardware. Even if it doesn't get to the same performance levels right away as the closed hardware is at, as long as it is "good enough" for most situations and works right out of the box I think people might buy it. I guess we'll see
      • That was the point of the original journal entry, though, no? That "good enough" wasn't quite good enough when it came to the gaming situations. That is what the proprietary vendors make money on. People are in a constant chase to keep up with the latest and greatest. Opengraphics.org has promise. Only time will tell if they will be able to create a viable business in an environment which, currently, is ted by proprietary players who not only have the majority of cash but also have the poli

If you are good, you will be assigned all the work. If you are real good, you will get out of it.