Announcement

Collapse
No announcement yet.

Intel Sandy Bridge Graphics Haven't Gotten Faster In Recent Years

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

  • #11
    Originally posted by Hi-Angel View Post
    And how does it explain the absence of performance improvement over the years?
    Hardware limit already reached all those years ago.

    Though to verify it that's really true, a comparison with Windows would be nice.

    Comment


    • #12
      Originally posted by wargames View Post
      The reason the performance has not improved anything over the years is pretty obvious: they want to buy their new shit.
      It is a SW problem, not HW one, more about how Ubuntu LTS is maintained.

      Originally posted by wargames View Post
      I'm waiting for ZEN!
      Yes, Intel needs a bit of competition from this direction too.

      Comment


      • #13
        Originally posted by Hi-Angel View Post
        And how does it explain the absence of performance improvement over the years?
        Well, how bigly should the performance improve over the years? Like the previous commenters mentioned, Sandy Bridge isn't particularly interesting. Ivy Bridge can probably reach OpenGL 4.1 soon, but Sandy is just too old.. Also, even if an overclocked (@ 4.x GHz) Intel i7 3770k somehow reached optimum perf on Linux, the performance would still be pretty modest. Those just lack the fast local RAM and shaders to do anything fancy.

        Comment


        • #14
          Originally posted by Hi-Angel View Post
          And how does it explain the absence of performance improvement over the years?
          The same way that there was an absence of performance improvement over the years for GMA3150 or whatever other turd card.
          The hardware itself is already used at near 100%. The hardware itself can't do much more than this and they said this when it was time to add fp64. Apart from wasting time polishing a turd, there isn't much else you can do.

          In this case, Intel and their mercenaries are focusing their efforts on hardware that actually needs it, like their newer iGPUs, and even Ivy Bridge is getting some love every now and then as the hardware in there is better.

          Comment


          • #15
            unapproved post for Hi-Angel , saying what others said already, anyway.

            Comment


            • #16
              Gusar , caligula wouldn't for example use of jemalloc make it faster? I do understand that the hardware old, but I honestly doubt that a way to improve performance for them doesn't exist anymore.

              Comment


              • #17
                Originally posted by Hi-Angel View Post
                Gusar , caligula wouldn't for example use of jemalloc make it faster? I do understand that the hardware old, but I honestly doubt that a way to improve performance for them doesn't exist anymore.
                That would be polishing a turd, i.e. not enough boost to make it worthwhile.

                Really, it's a DX10 SM4.1 GPU that on Windows exposes only OpenGL 3.1 (Linux has OpenGL 3.3), it lacks hardware for tessellation, various shader model 5 features, has no OpenCL support whatsoever.

                Unless you enable more features having even a 20% performance improvement is irrelevant, as that iGPU is very very likely relegated to "desktop compositing only" since a long time anyway.

                Comment


                • #18
                  I would be happy if stuck on render/blitter ring bug was resolved that is going on for years!

                  Comment


                  • #19
                    Originally posted by starshipeleven View Post
                    That would be polishing a turd, i.e. not enough boost to make it worthwhile.

                    Really, it's a DX10 SM4.1 GPU that on Windows exposes only OpenGL 3.1 (Linux has OpenGL 3.3), it lacks hardware for tessellation, various shader model 5 features, has no OpenCL support whatsoever.

                    Unless you enable more features having even a 20% performance improvement is irrelevant, as that iGPU is very very likely relegated to "desktop compositing only" since a long time anyway.
                    You're watching too globally. If you have had a notebook with such a hardware, would you just throw it away because "the GPU is a turd"? I didn't have SandyBridge, but I have an older one (with X3100 GPU), and though I'm not using it anymore, I'd be pretty much glad if once upon a time for I need it, I'd discover the driver got better.

                    What's the usecase? Well, e.g. to entertain a girlfriend whilst I'm busy.

                    Comment


                    • #20
                      Originally posted by starshipeleven View Post
                      That would be polishing a turd, i.e. not enough boost to make it worthwhile.

                      Really, it's a DX10 SM4.1 GPU that on Windows exposes only OpenGL 3.1 (Linux has OpenGL 3.3), it lacks hardware for tessellation, various shader model 5 features, has no OpenCL support whatsoever.

                      Unless you enable more features having even a 20% performance improvement is irrelevant, as that iGPU is very very likely relegated to "desktop compositing only" since a long time anyway.
                      I'm also pretty sure that users of Sandy Bridge graphics aren't holding their breath waiting for few FPS improvements in their favorite games. They've switched to discrete GPUs long ago if they're into gaming. And they're not switching back to iGPU if it gets any faster. I can see why someone would want to make it work instead of making fast. Having OpenGL 4.1 would be a really nice addition for normal desktop apps in case they ever start to require new OpenGL features. For instance it would be shame if LibreOffice didn't render bar plots anymore due to missing OpenGL support. The performance is secondary in such use.

                      Comment

                      Working...
                      X