Announcement

Collapse
No announcement yet.

Linux 3.5 Can Massively Boost AMD Radeon Graphics

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

  • #46
    Originally posted by Qaridarium View Post
    then why you don't use Windows7 on your "WINDOWS/Closed-Source-only" hardware?
    its a APU-FAKE if you use "Linux" because if you use a intel HD4000 graphic based INTEL-APU you get a much better "DEFAULT" result because the intel hardware is not "WINDOWS-ONLY/Closed-Source only"!
    AMD fail on your hardware on (Opensource) Linux thats the real point.

    o well in other words AMD FAIL to do Opensource drivers for there APU-Hardware.
    Well, Intel's competing product to the AMD C-Series iGPU is not the HD4000. It's the non working, IP-encumbered PowerVR SGX5 (AKA Intel GMA36xx). This abomination fits your description of a Fake-APU much more than anything AMD has ever released, at least when it comes to Open-Source Linux support.

    Comment


    • #47
      Originally posted by Qaridarium View Post
      then why you don't use Windows7 on your "WINDOWS/Closed-Source-only" hardware?

      As you suggested I think it's better for me to stick to Win7 on the crappy c-50 APU. And i'm really pleased with linux performance on Intel HD 3000 on my H61 mobo + SB Core i3 built.
      Of course, I'll keep visiting phoronix in the hope that AMD will someday release mature open source drivers as Intel does. Just a hope!

      Comment


      • #48
        Originally posted by manmath View Post
        As you suggested I think it's better for me to stick to Win7 on the crappy c-50 APU. And i'm really pleased with linux performance on Intel HD 3000 on my H61 mobo + SB Core i3 built.
        Of course, I'll keep visiting phoronix in the hope that AMD will someday release mature open source drivers as Intel does. Just a hope!
        AMD C-50 is a perfect little chip for a netbook. I love it and it uses much less resources than e.g. Intel-Atom if playing a video.

        I made some benchmarks here a while ago:
        http://openbenchmarking.org/result/1...AR-AMDC50PTS59

        Comment


        • #49
          Originally posted by disi View Post
          AMD C-50 is a perfect little chip for a netbook.
          http://openbenchmarking.org/result/1...AR-AMDC50PTS59
          Very very right. I've the similar experience on Windows 7. But on linux it's too much problematic. That's why I'm running only Win7 on it, and built a Linux machine with Intel h61 mobo and SB Core i3 processor, for my office needs.

          Comment


          • #50
            Originally posted by ceage View Post
            Well, Intel's competing product to the AMD C-Series iGPU is not the HD4000. It's the non working, IP-encumbered PowerVR SGX5 (AKA Intel GMA36xx). This abomination fits your description of a Fake-APU much more than anything AMD has ever released, at least when it comes to Open-Source Linux support.
            LOL FAIL intel FAIL to...

            Comment


            • #51
              Originally posted by manmath View Post
              As you suggested I think it's better for me to stick to Win7 on the crappy c-50 APU. And i'm really pleased with linux performance on Intel HD 3000 on my H61 mobo + SB Core i3 built.
              Of course, I'll keep visiting phoronix in the hope that AMD will someday release mature open source drivers as Intel does. Just a hope!
              If you don't care about using the catalyst proprietary driver on Windows, you could also use that driver on Linux to get nearly the same performance.

              Comment


              • #52
                Originally posted by manmath View Post
                On the same device Windows 7 handles 3d works and hd 1080p quite well. But linux with open source graphics simply fails to do that. I tried with catalyst, it works way better than the open source graphics, but it's too complicated in the sense i had to install catalyst drivers, vaapi-wrapper and then some workarounds in the media players. I thought open source graphics will be an easy way.
                What Q is trying to say is that the current open source graphics drivers run the GPU at whatever clocks the VBIOS set at startup unless you manually choose a different power profile. On discrete GPUs the default clocks are usually pretty high (full speed on older cards) so you get decent performance, but on APUs in laptops the default clocks are very low so default performance is low as well. I expect you'll see improvements in that regard as work on the drivers continues.

                Open source graphics drivers are "easier" in the sense that distros can integrate them for you, and test to make sure all the bits work together. They aren't easier overall, it's just that someone else can do the work for you *if* the drivers, libraries and players in the distro do everything you want. If you have to update the open source graphics drivers then its more of a wash - either easier or harder than installing proprietary drivers depending on what you are comfortable with.

                Originally posted by manmath View Post
                Just one puzzle, why does AMD releases such bad/shoddy open source drivers that are no where stand close to the proprietary ones.
                There are three different approaches to open source drivers among the major vendors. NVidia releases only proprietary drivers; Intel releases only open source drivers; we release proprietary drivers *and* work with the X.org developers on open source drivers.

                We don't "release" the open source drivers ourselves, other than an initial release of code and programming information to get the chip working and give the community (including our devs) a starting point to work on in public repositories. Also agd5f was the maintainer of the radeon X driver (xf86-video-ati, one key part of the open source driver stack) before joining AMD and he is continuing to manage releases of that driver.

                When we re-started support for open source graphics drivers a few years ago there were 4 generations of unsupported hardware to "catch up" on, with new hardware generations arriving every year. As a consequence, open source graphics driver support for new GPUs arrived quite a bit later than Catalyst support and there are still some gaps in functionality. It's probably fair to say that power management has become the most significant gap as a result of improvements in other areas, so we're going to try to push that ahead some more.

                Going forward, we expect that the next generation of GPUs should be the first where we are sufficiently caught up and have enough developers to provide launch-time support with the open source drivers. Each of the APU releases (Ontario, Llano and now Trinity) *have* had launch time support in the open source drivers, however, since their integrated GPUs were based on discrete GPU hardware which already had some support.
                Last edited by bridgman; 06-08-2012, 07:54 AM.

                Comment


                • #53
                  Originally posted by manmath View Post
                  Good to know, but I'm still skeptic about the OSS drivers performance when compared with catalyst drivers. Would somebody please tell me how it'll stand in performance in comparison to the catalyst drivers? Also please somebody tell does it improves ATI HD 6250 (AMD C-50 APU) performance? If so, how much?
                  I don't think this particular fix would improve performance on the C-50, although other work in progress (tiling, HyperZ among others) probably will.

                  Running at the same clocks, typical 3D performance of the open source drivers seems to be running around 40% of Catalyst right now, ranging from maybe 25% (Warsow) to 55%. These numbers should continue to improve over time, of course. If you look at r5xx, for example, the open source driver is running around 80% as fast as Catalyst, depending on the specific test, although that driver (r300g) also has a more sophisticated shader compiler than the one in r600g today.

                  EDIT - Great timing -- Michael just published a new benchmark on the older hardware : http://phoronix.com/forums/showthrea...053#post267053
                  Last edited by bridgman; 06-08-2012, 08:37 AM.

                  Comment


                  • #54
                    firmware

                    I've already tried building and installing 3.5-rc1 already, but I have some missing firmware. Can someone tell me where I can find this firmware?

                    ARUBA_rlc.bin
                    ARUBA_me.bin
                    ARUBA_pfp.bin
                    VERDE_rlc.bin
                    VERDE_mc.bin
                    VERDE_ce.bin
                    VERDE_me.bin
                    VERDE_pfp.bin
                    PITCAIRN_rlc.bin
                    PITCAIRN_mc.bin
                    PITCAIRN_ce.bin
                    PITCAIRN_me.bin
                    PITCAIRN_pfp.bin
                    TAHITI_rlc.bin
                    TAHITI_mc.bin
                    TAHITI_ce.bin
                    TAHITI_me.bin
                    TAHITI_pfp.bin

                    Comment


                    • #55
                      http://people.freedesktop.org/~agd5f/radeon_ucode/

                      Comment


                      • #56
                        You only need them if you have a Radeon GPU of the Aruba, Verde, Pitcairn or Tahiti family: in short, a 7000 series card. Else, you can ignore the message about missing firmware.

                        If you really need it, the firmware is here:
                        http://git.kernel.org/?p=linux/kerne...radeon;hb=HEAD

                        Comment


                        • #57
                          Originally posted by AlbertP View Post
                          You only need them if you have a Radeon GPU of the Aruba, Verde, Pitcairn or Tahiti family: in short, a 7000 series card. Else, you can ignore the message about missing firmware.

                          If you really need it, the firmware is here:
                          http://git.kernel.org/?p=linux/kerne...radeon;hb=HEAD
                          I have a kernel panic on boot related to the radeon drivers... I was assuming it was due to missing firmware. Perhaps the crash is unrelated to the firmware, we'll see once I try again (probably tonight after work).

                          Comment


                          • #58
                            Hello,

                            What are the remaining tasks to improve clocks profiles switching? Is it only bugs fix specificaly to some configuration/VBIOS/GPU model?

                            I'm asking this because with my RV770 I can switch from default to mid, but when I switch to low I "lose" my display. mid and low have the same voltage, only frequences changes. So I guess there is a workaround to implement, the Windows Catalyst uses the low profile.

                            How can I get some information useful to understand what is wrong when switching to low profile?

                            Comment


                            • #59
                              Originally posted by AlbertP View Post
                              If you don't care about using the catalyst proprietary driver on Windows, you could also use that driver on Linux to get nearly the same performance.
                              But it's not as straight forward. I tried once on debian squeeze, getting xvba-va-driver and some other stuff was very painful. Anyways, for now I'm going to stick with Win7, may be i'll check back sometime for the opensource driver.

                              Thanks all!

                              Comment


                              • #60
                                A lot of people are feeling left behind

                                Dear Phoronix,


                                I have this one little thing to say here that might be a good thing to consider once in a while:

                                For a lot of people, FGLRX won't work. But that is fine for us, because of course we have you and your awesome open source drivers - Thank you for this!!

                                But:

                                For some people, just some, even your awesome radeon driver does not work properly - Sometimes graphics are glitchy to a point, where it becomes unsuable, sometimes it seems impossible to properly set up the driver (e.g. on sandy- and ivy-machines). And sometimes, the radeon driver does not at all seem to be able to execute proper power-saving routines. (And don't even get started about WINE and radeon.)

                                I myself am not to much of a hot-shot when it comes to technical things - but I do know, that for the almost 3 years since I started to use linux now, the problems with AMD-cards have always been the same: Incompatibility (with everything) and power saving (battery life and noise)

                                Look, I know it must be a real pain to get things done when the actual manufacturer shows little support for an open source project. BUT: In these years I have gotten to know some people, who just don't believe anymore they will EVER see reliable Linux support for their AMD hardware. It's the known "it-works-for-some-but-not-for-all" effect - Some get a more or less good 3d acceleration out of radeon - and some don't.

                                But instead of closing this gap, it keeps on growing ever larger! Of course, everybody is happy when radeon keeps up the pace of performance with fglrx. But this doesn't help people who don't get any acceleration at all! So, we're all happy about huge performance increases with kernel 3.5 . But some people are still completely excluded from those benefits.

                                People are really getting alienated - not just by you, but by the whole linux-community at large.

                                For example: Starting with Torvalds' statements on the new kernels, over to the Ubuntu-Team, over to the WINE-community, over to phoronix, people have been keeping on claiming: "Finally - Sandybridge and Ivybridge support is working now!" - and have thus been turning to other issues.

                                Truth is: For a lot of machines, it doesn't work at all.

                                Truth is: If you've got an AMD card in you machine and you're not one of the lucky ones where it works out-of-the-box, you're pretty much screwed. VESA for you it is.

                                And: You won't get any help, because the people from your OS, say, Ubuntu are saying: "What do you want? - It works - it's not our fault", while torvalds is saying "What do you want? - It works - it's not my fault", while phoronix are saying "What do you want? - It works - it's not our fault" while the guys from WINE are saying "What do you want? - It works - it's not our fault".... you get the idea, I think.

                                So please - please help make it work, before you make it work fast.

                                Thank you!

                                Comment

                                Working...
                                X