Announcement

Collapse
No announcement yet.

People incorrectly assume that AMD drivers suck. They don't.

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • People incorrectly assume that AMD drivers suck. They don't.

    From a post taken of Slashdot...

    You are behind the times, and should really be firing your complaints at Nvidia. For the last couple of years I've used ATI cards for GL development exclusively. Unlike Nvidia cards they actually implement the GL spec to the letter. With Nvidia cards you can pretty much call any old combination of GL functions, and something will appear on screen. They never fail! This is a problem because you never find out errors in your GL code until after you've shipped the product. With ATI, if you pass an invalid arg, or call a method at the wrong time, they will generate the correct error. This sadly leads to a situation where a developer uses an NVidia card for development, ships, and then it won't run on ATI or Intel cards. The upshot is that people incorrectly assume that ATI drivers suck. They don't. Nvidia drivers are the ones that suck!
    You are behind the times, and should really be firing your complaints at Nvidia. For the last couple of years I've used ATI cards for GL development exclusively. Unlike Nvidia cards they actually implement the GL spec to the letter. With Nvidia cards you can pretty much call any old combination of G...


    I seen posts and blogs about this issue before, stuff written for Nvidia hardware sucks on AMD/Intel hardware cause Nvidia cares not about specs and correctness, just speed and hacks in their OpenGL stack.


    Thoughts?

  • #2
    Example in Gnome Shell code: http://ati.cchtml.com/show_bug.cgi?id=432#c14

    Comment


    • #3
      Well, while that is certainly a fact, I've found that Mesa is even more strict. As a matter of fact I have in my hands an app that renders correctly on r600g but fails several areas on fglrx.

      Comment


      • #4
        Originally posted by Rallos Zek View Post
        I seen posts and blogs about this issue before, stuff written for Nvidia hardware sucks on AMD/Intel hardware cause Nvidia cares not about specs and correctness, just speed and hacks in their OpenGL stack.


        Thoughts?
        That could be true. But if I have to choose between a driver that follows specs but doesn't work properly and a driver that uses various hacks but does work and is stable I'd definitly choose the later. Who cares if the driver is "correct" if it crashes the computer?

        Comment


        • #5
          Maybe amd should send free cards to kde and gnome devs to optmize the code path. I really think amd should do more than they did before. Most likely amd devs are forced to use ubuntu at work and when that works they are happy. At home fglrx devs play xbox 360 (with amd chip) and do not use Linux most likely

          Comment


          • #6
            Originally posted by RussianNeuroMancer View Post
            This is actually also a good example of why good bug reporters are required.... didn't get a fix found until after someone got a reasonable back trace for amd to go off of.


            Originally posted by nej_simon View Post
            That could be true. But if I have to choose between a driver that follows specs but doesn't work properly and a driver that uses various hacks but does work and is stable I'd definitly choose the later. Who cares if the driver is "correct" if it crashes the computer?
            That's the type of logic that produced the Reliant Robin, and kept that going for over 20 years. Although, It's also an important example of why having specifications and following these specifications is key.

            Wikipedia: Reliant Robin
            Youtube.com - Top Gear

            Originally posted by Kano View Post
            Maybe amd should send free cards to kde and gnome devs to optmize the code path. I really think amd should do more than they did before. Most likely amd devs are forced to use ubuntu at work and when that works they are happy. At home fglrx devs play xbox 360 (with amd chip) and do not use Linux most likely

            I'd actually tend to agree. That, or at least make sure that one or two of the core devs has an amd card at all times and they work with the dev team by submittin bug reports... although, more likely, they will issue code to prevent these bugs from appearing in the first place.

            Comment


            • #7
              Originally posted by Rallos Zek View Post
              From a post taken of Slashdot...



              You are behind the times, and should really be firing your complaints at Nvidia. For the last couple of years I've used ATI cards for GL development exclusively. Unlike Nvidia cards they actually implement the GL spec to the letter. With Nvidia cards you can pretty much call any old combination of G...


              I seen posts and blogs about this issue before, stuff written for Nvidia hardware sucks on AMD/Intel hardware cause Nvidia cares not about specs and correctness, just speed and hacks in their OpenGL stack.


              Thoughts?
              Wait wait wait, the reason why I don't buy ATI/AMD is much more simple. You see, in 2008, when trying Ubuntu 8.04 perfectly working with nvidia's 6800 card, on a HD4670 - the catalyst has simply frozen my desktop and it constantly froze the system if I switched to TTY. It was a clean install. In fact, the single working thing was OpenGL-based OpenArena. But later on came AMD' opensource story about plans for a niche driver of 3rd class citizenship. Right now I'm looking at Intel, which does the stuff right, because AMD's opensource is still underperforming and those who disagree are tossed back into proprietary driver area. Im in that area with nvidia already, works stable from here.

              By the way, you should be asking programmers who actually work with OpenGL about quality of drivers. Ask Carmack for example.

              I excuse myself in front of ATI/AMD developers and hackers there, because the paragraph above applies only to the "official" management.

              Comment


              • #8
                You confuse testing suits and debug flags/behaviour with release and production versions. Production versions should continue if its possible, while debug should complain on every possible flaw. 80-year-old grannies with ubuntu ain't engineers!

                Comment


                • #9
                  Is there a compliance test for opengl does anybody know of?

                  kind of like acid3 ( http://acid3.acidtests.org/ )

                  P.S. gnome-shell is awful!

                  Comment


                  • #10
                    Originally posted by D0pamine View Post
                    Is there a compliance test for opengl does anybody know of?

                    kind of like acid3 ( http://acid3.acidtests.org/ )

                    P.S. gnome-shell is awful!
                    The only "affordable" compliance test end users can try is called piglit... see it here: http://cgit.freedesktop.org/piglit/

                    Comment

                    Working...
                    X