Announcement

Collapse
No announcement yet.

Intel Sandy Bridge Linux Graphics? It's A Challenge

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

  • #31
    Originally posted by spykes View Post
    So in the end, we will have to wait for Q3/Q4 of 2011 in order to see a "proper" support of Sandy Bridge packed in our distribution.
    Thankfully they have started the Linux support a year ago...
    Gentoo Linux users probably do not have to wait that long. The source code is available now, so they just need to switch to instruct the package manager to install it, which might be somewhat involved, but it will become easier as more Gentoo Linux users obtain Sandy Bridge processors and write ebuilds for the newer software, assuming that they have not written them already.

    Comment


    • #32
      Sandybridge is out, but no Gallium support or support expected in later drivers...makes me wonder. Does Intel actually plan to some day develop and optimize their open source drivers to the point of making them faster than what Gallium is capable of? What other reason would they have for not wanting to begin transitioning to Gallium?

      Comment


      • #33
        Currently Gallium sufferes from vaious bottlenecks hurting performance especially on weaker IGPs utilizing shared memory. For instance the mobile Radeon HD4250 does much less FPS in OpenGL games than with the classic mesa stack. With faster GPUs is seems to be the opposite in some cases. My best guess ist that Intel avoids Gallium at this time to maximize the performance of their IGPs under Linux - just my two cents

        - saski

        Comment


        • #34
          To be clear, a hardware problem that has existed since Clarkdale is preventing accelerated playback of 23.97 FPS video without stuttering on Sandy Bridge's graphics?

          Comment


          • #35
            Originally posted by popper View Post
            No..i think its just a lack of time before they fix that programming error
            It is not a "programming error". It is a hardware limitation.

            Comment


            • #36
              Originally posted by mattst88 View Post
              To be clear, a hardware problem that has existed since Clarkdale is preventing accelerated playback of 23.97 FPS video without stuttering on Sandy Bridge's graphics?
              Yes, that is the case. Intel has had a loooooooooooong time to fix that issue with their hardware.

              Comment


              • #37
                Originally posted by deanjo View Post
                Yes, that is the case. Intel has had a loooooooooooong time to fix that issue with their hardware.
                I don't guess there are any workarounds other than an extra frame every 40* seconds?

                * (24 frames/sec - 23.976 frames/sec) * 40 sec = ~1 frame.

                Comment


                • #38
                  Originally posted by mattst88 View Post
                  I don't guess there are any workarounds other than an extra frame every 40* seconds?

                  * (24 frames/sec - 23.976 frames/sec) * 40 sec = ~1 frame.
                  Sure there is, get an nvidia.

                  Comment


                  • #39
                    Indeed.

                    The difficulty of that workaround is that one has to spend more money on an otherwise unnecessary graphics card that would replace the otherwise OK integrated graphics provided on Sandy Bridge CPUs.

                    Comment


                    • #40
                      Originally posted by mattst88 View Post
                      Indeed.

                      The difficulty of that workaround is that one has to spend more money on an otherwise unnecessary graphics card that would replace the otherwise OK integrated graphics provided on Sandy Bridge CPUs.
                      A $30 dollar graphics card is a small price to pay.

                      Comment


                      • #41
                        Originally posted by deanjo View Post
                        It is not a "programming error". It is a hardware limitation.
                        well i did put a there ;D
                        but still... is it a case of a control 'bit' being set/unset on or off wrongly, or a 'real hardware errata' that they know about and have not fixed, did anyone bother to report this known error on intel-gfx@lists.freedesktop.org for instance ?

                        Comment


                        • #42
                          Originally posted by mattst88 View Post
                          I don't guess there are any workarounds other than an extra frame every 40* seconds?

                          * (24 frames/sec - 23.976 frames/sec) * 40 sec = ~1 frame.
                          According to that Anandtech article, Intel did come up with a workaround in their drivers that can change the framerate to 23.970. That comes out to 1 frame every 160 seconds.

                          Comment


                          • #43
                            Originally posted by mattst88 View Post
                            To be clear, a hardware problem that has existed since Clarkdale is preventing accelerated playback of 23.97 FPS video without stuttering on Sandy Bridge's graphics?
                            which strikes me as odd, given that Clarkdale uses a dual-core Westmere and sticks it next to the 45nm Intel GMA die, where as Sandy bridge is totally new and in the core and uses the new rings tied the the L3.

                            that implies program error somewhere not hardware errata, someone should perhaps ask for clarity and post a test case to prove it on intel-gfx@lists.freedesktop.org as we are just assuming right now it seems!

                            Comment


                            • #44
                              Originally posted by popper View Post
                              which strikes me as odd, given that Clarkdale uses a dual-core Westmere and sticks it next to the 45nm Intel GMA die, where as Sandy bridge is totally new and in the core and uses the new rings tied the the L3.

                              that implies program error somewhere not hardware errata, someone should perhaps ask for clarity and post a test case to prove it on intel-gfx@lists.freedesktop.org as we are just assuming right now it seems!
                              From Anandtech, it's a hardware problem that Intel knows about.

                              The limitation is entirely in hardware, particularly in whatís supported by the 5-series PCH (remember that display output is routed from the processorís GPU to the video outputs via the PCH). One side effect of trying to maintain Intelís aggressive tick-tock release cadence is thereís a lot of design reuse. While Sandy Bridge was a significant architectural redesign, the risk was mitigated by reusing much of the 5-series PCH design. As a result, the hardware limitation that prevented a 23.976Hz refresh rate made its way into the 6-series PCH before Intel discovered the root cause.

                              Intel had enough time to go in and fix the problem in the 6-series chipsets, however doing so would put the chipset schedule at risk given that fixing the problem requires a non-trivial amount of work to correct. Not wanting to introduce more risk into an already risky project (brand new out of order architecture, first on-die GPU, new GPU architecture, first integrated PLL), Intel chose to not address it this round, which is why we still have the problem today.

                              Comment


                              • #45
                                Originally posted by deanjo View Post
                                Correct me if I'm wrong but IIRC Media Espresso doesn't actually do the actual encoding on the GPU, that is still done on the CPU. What is done on the GPU is the filtering, resizing etc the same as TMPEnc does.
                                They have a special version that enables HW decode & encode on Intel Sandy Bridge. Likewise, the CUDA version uses the new NVIDIA component. Hence, quality is not that good.

                                Some tests:
                                http://www.anandtech.com/show/4083/t...2100-tested/9#
                                http://www.legitreviews.com/article/1506/8/
                                http://static.pcinpact.com/images/bd...espresso-6.jpg

                                Comment

                                Working...
                                X