Announcement

Collapse
No announcement yet.

amd-staging-4.6 for Fedora 24 (AMDGPU)

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

  • #31
    Originally posted by kgonzales View Post
    Mystro256 Here are some updates:

    1) The newest kernel build (kernel-4.6.7-901.amd.16082016.fc24.x86_64) will boot on my system without issues.
    2) I have load your bits (your Mesa build, your X.org AMDGPU driver, your AMD staging kernel), and the system is using both your kernel and the X.org AMDGPU driver.
    3) The performance of this stack is less good than the griever Mesa git repo, by about 3-5%. I have more graphical artifacts with your stack (but to be honest, the open drivers have graphical artifacts, period) and also more screen tearing.

    So. Your stuff works, but as you mentioned in an earlier post... this does not appear to be improving performance. And with the patch earlier mentioned improved performance in the radeon driver for muxless systems with SI/VI cards improving DRI_PRIME performance (literally my top use case), I am moving back to the griever repo. However, if you have something new you are interested in having me test, please let me know! I have also learned my way around the advanced commands for DNF during this exercise, and its now WAY easier to switch between repos.
    Good to know you got something out of it.

    As I said, with older cards, the only real advantage to amdgpu is if you're looking to use any possible new functionality with your card, such as freesync (do laptops even support this?). As of now, there's not a lot of reason to switch over with Fedora, but this may change when they port the hybrid driver to RHEL, which could mean a functioning driver for Fedora. I've tried repackaging it a few times from Ubuntu, no luck, but RHEL is much more similar stack wise to Fedora, so we may have luck there.

    Comment


    • #32
      Originally posted by bridgman View Post
      It's getting there... which of those two is "the AMDGPU package" ? Or does it depend on context ?
      Probably. I think "the AMDGPU package" is the X.org AMDGPU driver that Mystro256 provided. Now... I don't know what that is exactly (other than it exists) but I guess there is open source AMDGPU driver for X.org now, instead of using the "radeon" X.org driver. At least that is what it was doing on my system.

      Comment


      • #33
        Originally posted by Mystro256 View Post

        Good to know you got something out of it.

        As I said, with older cards, the only real advantage to amdgpu is if you're looking to use any possible new functionality with your card, such as freesync (do laptops even support this?). As of now, there's not a lot of reason to switch over with Fedora, but this may change when they port the hybrid driver to RHEL, which could mean a functioning driver for Fedora. I've tried repackaging it a few times from Ubuntu, no luck, but RHEL is much more similar stack wise to Fedora, so we may have luck there.
        Lenovo created one Y700-15ACZ model with a M380 video card and a Freesync panel. But it's hard to find that Bigfoot and Nessie combined. It was sold at Best Buy only, and it appears that Best Buy has sent the Y700's to the outlets and refurb stores. HP also created one:



        Now if Freesync worked with an external monitor off of a laptop... that would be sexy and worth testing again. bridgman Any ideas here?

        If Airlie continues driving the open source Vulkan support for AMDGPU, then that would be another interesting thing to test.



        Comment


        • #34
          Originally posted by kgonzales View Post
          Probably. I think "the AMDGPU package" is the X.org AMDGPU driver that Mystro256 provided. Now... I don't know what that is exactly (other than it exists) but I guess there is open source AMDGPU driver for X.org now, instead of using the "radeon" X.org driver. At least that is what it was doing on my system.
          Still a bit confused. IIRC you were the one using the term "the AMDGPU package" so shouldn't you know exactly what it is ? Not necessarily the contents, but where you got it from / what you downloaded / installed etc.. ?
          Test signature

          Comment


          • #35
            Originally posted by bridgman View Post

            Still a bit confused. IIRC you were the one using the term "the AMDGPU package" so shouldn't you know exactly what it is ? Not necessarily the contents, but where you got it from / what you downloaded / installed etc.. ?
            I think he/she's confusing things.. kgonzales is referring to my build of the amd staging kernel plus my build of the amdgpu xorg driver as "the AMDGPU package".

            The kernel is a build from this commit plus the redhat kernel patch set and kernel 4.6.7 patch set (from kernel,org).

            The xorg driver is a build from this commit by Michel Dänzer.

            Comment


            • #36
              Originally posted by kgonzales View Post
              Mystro256 Here are some updates:
              ...
              And with the patch earlier mentioned improved performance in the radeon driver for muxless systems with SI/VI cards improving DRI_PRIME performance (literally my top use case), I am moving back to the griever repo. However, if you have something new you are interested in having me test, please let me know!
              ...
              The DRI_PRIME patch, are you referring to this?

              Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite


              If so, it's a patch for radeonSI, meaning it should benefit amdgpu + mesa as well, though I haven't actually looked at the source yet. Note that it hasn't been merged into mesa-git yet, or since i last checked, so I would hold tight on that.

              I'll be away for the next few days, but next week I plan to update a bunch of things in my repos:
              - LLVM 3.9, if it gets released on time
              - Mesa to the latest accepted radeon and radeonsi fixes, if any.
              - And I just noticed they changed some code for tonga in staging, so I'll get this into my next build of the kernel.

              Check back in around Wednesday (should be built by then) and let me know if you get a chance to test it

              Comment


              • #37
                Originally posted by Mystro256 View Post

                The DRI_PRIME patch, are you referring to this?

                Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite


                If so, it's a patch for radeonSI, meaning it should benefit amdgpu + mesa as well, though I haven't actually looked at the source yet. Note that it hasn't been merged into mesa-git yet, or since i last checked, so I would hold tight on that.
                Yeah that is the patch I was referring to. I'll certainly wait for it as I expect some great things there. I also saw this patch being developed:

                Marek Olšák has recently sent word to the AMD mailing list that they have found a reason for some games performing poorly using Mesa. Another developer noted that a patch is already in progress.


                I have actually seen this happen in Civ5 in the late game. Graphical freezes where everything just stops for a second or two before moving forward. Would love to see it fixed.

                Originally posted by Mystro256 View Post

                I'll be away for the next few days, but next week I plan to update a bunch of things in my repos:
                - LLVM 3.9, if it gets released on time
                - Mesa to the latest accepted radeon and radeonsi fixes, if any.
                - And I just noticed they changed some code for tonga in staging, so I'll get this into my next build of the kernel.

                Check back in around Wednesday (should be built by then) and let me know if you get a chance to test it
                I'll be away myself for most of next week but I will certainly test it next weekend. Just drop a note in this thread once it's ready and I will get it loaded up.

                Comment


                • #38
                  Originally posted by bridgman View Post

                  Still a bit confused. IIRC you were the one using the term "the AMDGPU package" so shouldn't you know exactly what it is ? Not necessarily the contents, but where you got it from / what you downloaded / installed etc.. ?
                  bridgman, if you are still confused, as far as I can see from following this conversation, "the AMDGPU package" is the xorg-x11-drv-amdgpu Fedora rpm from Mystro256's copr here http://copr-fe.cloud.fedoraproject.o...6/polaris-gfx/

                  Comment


                  • #39
                    Hi Mystro256,
                    Thank you for providing recent kernel with CIK support enabled. Here is the result with AMD A10-7400P Kaveri R6 (Possibly a 265X) using the boot parameter to disable radeon module. Hardware is a ASUS X550ZE laptop.
                    System used is Fedora 24 Design Suite based on Workstation running on Gnome Wayland.
                    Code:
                    $ dmesg | grep amdgpu
                    [    0.000000] Command line: BOOT_IMAGE=/vmlinuz-4.6.7-900.amd.16082016.fc24.x86_64 root=UUID=232c0ccf-375a-4e4d-a080-9986ea920e1c ro rhgb quiet LANG=en_CA.UTF-8 modprobe.blacklist=sp5100_tco amdgpu.exp_hw_support=1 modprobe.blacklist=radeon
                    [    0.000000] Kernel command line: BOOT_IMAGE=/vmlinuz-4.6.7-900.amd.16082016.fc24.x86_64 root=UUID=232c0ccf-375a-4e4d-a080-9986ea920e1c ro rhgb quiet LANG=en_CA.UTF-8 modprobe.blacklist=sp5100_tco amdgpu.exp_hw_support=1 modprobe.blacklist=radeon
                    [    1.931419] [drm] amdgpu kernel modesetting enabled.
                    [    1.946637] fb: switching to amdgpudrmfb from EFI VGA
                    [    1.947436] amdgpu 0000:00:01.0: VRAM: 1024M 0x0000000000000000 - 0x000000003FFFFFFF (1024M used)
                    [    1.947439] amdgpu 0000:00:01.0: GTT: 7493M 0x0000000040000000 - 0x00000002145807FF
                    [    1.947472] [drm] amdgpu: 1024M of VRAM memory ready
                    [    1.947474] [drm] amdgpu: 7493M of GTT memory ready.
                    [    2.048985] amdgpu 0000:00:01.0: amdgpu: using MSI.
                    [    2.049018] [drm] amdgpu: irq initialized.
                    [    2.049026] [drm] amdgpu: dpm initialized
                    [    2.052710] [drm] amdgpu atom DIG backlight initialized
                    [    2.053073] amdgpu 0000:00:01.0: fence driver on ring 0 use gpu addr 0x0000000040000010, cpu addr 0xffff880437989010
                    [    2.053150] amdgpu 0000:00:01.0: fence driver on ring 1 use gpu addr 0x0000000040000020, cpu addr 0xffff880437989020
                    [    2.053283] amdgpu 0000:00:01.0: fence driver on ring 2 use gpu addr 0x0000000040000030, cpu addr 0xffff880437989030
                    [    2.053349] amdgpu 0000:00:01.0: fence driver on ring 3 use gpu addr 0x0000000040000040, cpu addr 0xffff880437989040
                    [    2.053409] amdgpu 0000:00:01.0: fence driver on ring 4 use gpu addr 0x0000000040000050, cpu addr 0xffff880437989050
                    [    2.053473] amdgpu 0000:00:01.0: fence driver on ring 5 use gpu addr 0x0000000040000060, cpu addr 0xffff880437989060
                    [    2.053526] amdgpu 0000:00:01.0: fence driver on ring 6 use gpu addr 0x0000000040000070, cpu addr 0xffff880437989070
                    [    2.053583] amdgpu 0000:00:01.0: fence driver on ring 7 use gpu addr 0x0000000040000080, cpu addr 0xffff880437989080
                    [    2.053629] amdgpu 0000:00:01.0: fence driver on ring 8 use gpu addr 0x0000000040000090, cpu addr 0xffff880437989090
                    [    2.053723] amdgpu 0000:00:01.0: fence driver on ring 9 use gpu addr 0x00000000400000a0, cpu addr 0xffff8804379890a0
                    [    2.053775] amdgpu 0000:00:01.0: fence driver on ring 10 use gpu addr 0x00000000400000b0, cpu addr 0xffff8804379890b0
                    [    2.055539] amdgpu 0000:00:01.0: fence driver on ring 11 use gpu addr 0x0000000000f2fd30, cpu addr 0xffffc90001e38d30
                    [    2.055834] amdgpu 0000:00:01.0: fence driver on ring 12 use gpu addr 0x00000000400000d0, cpu addr 0xffff8804379890d0
                    [    2.055894] amdgpu 0000:00:01.0: fence driver on ring 13 use gpu addr 0x00000000400000e0, cpu addr 0xffff8804379890e0
                    [    3.262984] fbcon: amdgpudrmfb (fb0) is primary device
                    [    6.839475] amdgpu 0000:00:01.0: fb0: amdgpudrmfb frame buffer device
                    [    7.094807] [drm] Initialized amdgpu 3.3.0 20150101 for 0000:00:01.0 on minor 0
                    [   18.665275] audit: type=1130 audit(1471911351.094:64): pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=systemd-backlight@backlight:amdgpu_bl0 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
                    A few observations:
                    • The backlight issue happening on radeon module is gone when using amdgou module
                    • The consumption of battery seems improving as the result and the power management seems to work properly
                    • Noticeable improvement of performance in Gnome Wayland
                    • Without the boot parameter, slowdown occurred


                    The second Radeon card is Jet PRO [Radeon R5 M230] (SI card?) which remain unclaimed according to lshw command. It will be interesting to have that card enabled once the hybrid support get resolved.

                    I will keep the amd-stage-kernel copr repository enabled for a while. I think the amdgpu kernel module is working well for CIK and should be enabled by default on future kernel release.

                    Comment


                    • #40
                      Originally posted by debianxfce View Post
                      ...
                      I wish I could delete comments on my forums threads because the shit you spew isn't welcome here.
                      This is not an opinion thread, nor am I interested in hearing your opinion on AMD's business models.
                      Last edited by Mystro256; 23 August 2016, 09:20 AM.

                      Comment

                      Working...
                      X