Announcement

Collapse
No announcement yet.

TitaniumGL: A Faster Multi-Platform Graphics Driver Architecture?

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

  • #11
    I see there are some people who suspect cheating here (which is sad, given that no one has actually verified the accusations yet). It should be easy to find out for anyone with at least two different video cards. Just do a benchmark, then swap in the other card and see if the results change considerably.

    On another note, I've seen bits of the original source code myself. Believe me, most people wouldn't understand a thing. All functions and variables have crazy_long_non_english_naming.

    On a third note, I can actually imagine that TitaniumGL is faster than LLVMpipe, given that LLVMpipe probably tries to be reasonably accurate and emulate shaders. TitaniumGL probably makes a hell of a lot of shortcuts (which is a perfectly valid approach, if the unsued functions are not/seldomly used, why not leave them out?). Either way, more testing needed here.
    Last edited by ultimA; 09 March 2012, 12:45 PM.

    Comment


    • #12
      Originally posted by t.s. View Post
      Yep, +1. Hire him immediately!
      Who's gonna hire him?

      Comment


      • #13
        Unless you forced llvmpipe to GL 1.4 too, the games very likely took different paths, using heavier shaders on llvmpipe where the ability is there.

        Also different WM (unity vs unity 2d).

        Comment


        • #14
          Originally posted by ultimA View Post
          I see there are some people who suspect cheating here (which is sad, given that no one has actually verified the accusations yet). It should be easy to find out for anyone with at least two different video cards. Just do a benchmark, then swap in the other card and see if the results change considerably.

          On another note, I've seen bits of the original source code myself. Believe me, most people wouldn't understand a thing. All functions and variables have crazy_long_non_english_naming.

          On a third note, I can actually imagine that TitaniumGL is faster than LLVMpipe, given that LLVMpipe probably tries to be reasonably accurate and emulate shaders. TitaniumGL probably makes a hell of a lot of shortcuts (which is a perfectly valid approach, if the unsued functions are not/seldomly used, why not leave them out?). Either way, more testing needed here.
          the thing is, either llvm is failing (with multiple devs on it, whereas titanium, seems to be a one man show - maybe not, idk) or its "focusing on different things". so what do you think is the sad thing? people being suspicious to a suddenly appeared architecture they dont know about which is closed source is only rathional.
          either:
          a) the gallium devs are a bunch of morons having lost several years on developing some crap
          b) this guy is a genious, or
          c) we are comparing apples to pears here. leaving the question which we like better: apples or pears?

          or:
          d) im mixing something up. like gallium and llvm or something

          i wonder if this guy is referring to the same decisions about gallium that hold intel back from porting their drivers to gallium.

          PS.:
          e) michael needs money and invented some story to get clicks on the site and forum =D

          Comment


          • #15
            Originally posted by schmidtbag View Post
            personally i don't care if something like this is closed source, but the only reason why i'd like it to be open source is so people could compile it for other architectures - architectures that NEED it like PPC. the hardware limitations on this are pretty ridiculous tho.
            There are two kinds of people:
            - those who once got "owned" by proprietary blob and prefer opensource solution anyday (if it backed by company)
            - those who are on their way to be "owned" soon. They sometimes also call 1st group "opensource maniacs".

            Ok, the final application or front end, in extreme cases, CAN be closed source; or the driver, if it does the job, bundled exactly with the hardware, is not specific and security is not a factor.

            But library?! I would trust unigine binary or even DRM coders(as in rights mismanagement) way more than this library.

            This is completely wrong way to earn money. The guy should please really rethink his development method! Even opencore is way better.

            By the way 0access rootkit earns money exactly THIS way (opening windows with ads and autoclicking). So, this is graphical rootkit.

            Comment


            • #16

              Comment


              • #17
                Someone should try to run a quick piglit test on that.
                And compare with, say, llvmpipe.

                Comment


                • #18
                  I hadn't heard about this before but from TFA
                  TitaniumGL on Windows attempts to implement OpenGL functionality over Direct3D (i.e. translating OpenGL calls into Direct3D so that they can be executed by a host driver, which is similar to a few other software projects out there). This at least means GPU-based acceleration, but under Linux, BSD, and ReactOS the TitaniumGL library basically comes down to being a software-based rasterizer running on the CPU
                  would someone compare this architecture to the native opensource drivers (not llvmpipe)? I've got the distinct impression that this isn't a software raster based upon the results but simply just still a translator. We can't tell by looking at the code of course... but benchmarks will tell us

                  Comment


                  • #19
                    Originally posted by crazycheese
                    a) the gallium devs are a bunch of morons having lost several years on developing some crap
                    They're busy making sure that the capabilities of your hardware are taken advantage of. Personally I find that more important than software rendering performance in obsolete games.

                    Comment


                    • #20
                      Originally posted by crazycheese View Post
                      A library, that is closed source, fetching advertisements and translates opengl calls into proprietary directx?

                      Well that just sounds like a graphical rootkit!
                      But that's on Windows

                      Originally posted by ChrisXY View Post
                      Maybe a test for the actual image quality / rendering accuracy would be beneficial to the evaluation of the performance?
                      Good point, I didn't thought about it.

                      Originally posted by ultimA View Post
                      I see there are some people who suspect cheating here (which is sad, given that no one has actually verified the accusations yet). It should be easy to find out for anyone with at least two different video cards. Just do a benchmark, then swap in the other card and see if the results change considerably.

                      On another note, I've seen bits of the original source code myself. Believe me, most people wouldn't understand a thing. All functions and variables have crazy_long_non_english_naming.

                      On a third note, I can actually imagine that TitaniumGL is faster than LLVMpipe, given that LLVMpipe probably tries to be reasonably accurate and emulate shaders. TitaniumGL probably makes a hell of a lot of shortcuts (which is a perfectly valid approach, if the unsued functions are not/seldomly used, why not leave them out?). Either way, more testing needed here.
                      Or just test it with an Unichrome and mesa 8.0, since it was dropped (which means there is no way to call the mesa driver).

                      Comment

                      Working...
                      X