Announcement

Collapse
No announcement yet.

Trying Out Nouveau GPU Re-Clocking On Linux 3.16

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

  • Trying Out Nouveau GPU Re-Clocking On Linux 3.16

    Phoronix: Trying Out Nouveau GPU Re-Clocking On Linux 3.16

    With the Linux 3.16 kernel comes the ability to re-clock select NVIDIA GeForce GPUs when using the open-source, reverse-engineered Nouveau driver. Here's my first impressions with trying out this option to maximize the performance of NVIDIA graphics cards on open-source drivers.

    http://www.phoronix.com/vr.php?view=20609

  • #2
    The benchmarks will be interesting! Could you include (archived) OSS-driver results from similar-priced AMD cards for reference?

    Comment


    • #3
      Michael, Do you ever fear about hardware dying/breaking while doing these types of tests?

      I'll probably wait until its at least 90% reliable before testing it out myself.

      Comment


      • #4
        Originally posted by oleid View Post
        The benchmarks will be interesting! Could you include (archived) OSS-driver results from similar-priced AMD cards for reference?
        That would require he uses the same computer and setup, which with the way he seems to cycle through OSs, he probably doesn't have the same install anymore. Then again, I'm just assuming, I really have no fucking clue

        Comment


        • #5
          It's a shame the fan goes instantly at 100%, it can't be considered usable until proper fan control is done.

          Comment


          • #6
            Fans can be modded or put on hardware controllers

            Originally posted by Calinou View Post
            It's a shame the fan goes instantly at 100%, it can't be considered usable until proper fan control is done.
            If the only issue in reclocking was fan control, I'd wire the fans to a $5 rheostat from Radio Shack as a controller. I've got two of those on my current machine, one for the case fans, and one for the voltage-controlled fans on a big tower CPU cooler whose 135mm fans are only voltage controlled. Not enough control range and entirely too noisy, so I turn them down in normal use and full throttle them for video rendering jobs.

            Comment


            • #7
              Originally posted by Calinou View Post
              It's a shame the fan goes instantly at 100%, it can't be considered usable until proper fan control is done.
              Fan management is done, the bug has been found (the temperature couldn't be read at higher clocks) and fixed ( http://cgit.freedesktop.org/~darktam...cfe77b36c66ca6 ).

              Comment


              • #8
                It's probably better to make such tests on a laptop system (no fan, and as usual it's a good guess that some developers use laptops/thinkpads).

                Comment


                • #9
                  Some laptops have loud screeing fans

                  Originally posted by not.sure View Post
                  It's probably better to make such tests on a laptop system (no fan, and as usual it's a good guess that some developers use laptops/thinkpads).
                  Laptops, even my little netbook, do in fact have fans. Some of these tiny, small-diameter fans make a loud screeeing noise when full throttled

                  Comment


                  • #10
                    Originally posted by Luke View Post
                    Laptops, even my little netbook, do in fact have fans. Some of these tiny, small-diameter fans make a loud screeeing noise when full throttled
                    Yes, but they are not controlled by Nouveau

                    Comment


                    • #11
                      Originally posted by not.sure View Post
                      It's probably better to make such tests on a laptop system (no fan, and as usual it's a good guess that some developers use laptops/thinkpads).
                      Why do you mention ThinkPads as if they are totally different from laptops? Even if you were to say "laptops, specifically ThinkPads", I'm not sure why it's worth mentioning.

                      Comment


                      • #12
                        well, and there I was hoping that it would be immediately usable after activating

                        hope these kind of issues (graphics, garbled screen) can be sorted out rather quickly


                        anyways: great work and massive thanks to the devs - that it's at least in near reach now to be usable

                        Comment


                        • #13
                          Originally posted by not.sure View Post
                          It's probably better to make such tests on a laptop system (no fan, and as usual it's a good guess that some developers use laptops/thinkpads).
                          Yep. Sager NP7330 GFX 765m doesn't have it's own INDIVIDUAL fan, but it shares a connection with the CPU HS/fan via heatpipe.

                          The Sager NP8655(trying to remember model #, 15.6") w/GFX 780m has a separate HS/fan setup for the CPU and another for the GPU.

                          My old MSI GT725 w/4850 mobility had a shared setup like the 7330.

                          Pretty much any dGPU is going to have some kind of fan setup and NOT just rely on a heatsink. Mobile dGPUs run a bit warmer than their desktop counterparts, e.g. 780m is specwise, very similar to the GTX 670(which I have), not sure what the 765m corresponds to from the desktop... 660 or 650 maybe?(768 shaders) And the old 4850 was an oddity, same specs as the 4850 desktop variant(shader count) but w/lower core/mem clocks(the ONLY time that I can recall that current mobile GPUs had identical specs other than core/mem clocks as analogous desktop parts, didn't last long though, next iteration was back to the not quite matching similarly "named" desktop part).

                          Comment


                          • #14
                            So still no reclocking for the GTX 460... what a pain. The old performance_level interface worked fine. I hate it when they remove something that works to replace it with something that doesn't. I've been stuck at 50/135 MHz for several months, while I could set 675/1800 MHz without any problem until kernel 3.12 with "echo 3 > performance_level"

                            Comment


                            • #15
                              For that one solution is obvious

                              Originally posted by zequav View Post
                              So still no reclocking for the GTX 460... what a pain. The old performance_level interface worked fine. I hate it when they remove something that works to replace it with something that doesn't. I've been stuck at 50/135 MHz for several months, while I could set 675/1800 MHz without any problem until kernel 3.12 with "echo 3 > performance_level"
                              Install kernel 3.11 and use that for your normal work unless other changes gave a greater performance increase than the loss of reclocking cost you. Keep testing the new ones as they come out, looking for the new reclocking support. My netbook won't show the splash screen with kernel 3.15, so it is staying on 3.14 until either that gets fixed, a new kernel offers some other advantage like better video playback, or newer operating systems become incompatable with 3.14.

                              Comment

                              Working...
                              X