Announcement

Collapse
No announcement yet.

AMD Open-Source S.I. Botched, Hope For The Future

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

  • AMD Open-Source S.I. Botched, Hope For The Future

    Phoronix: AMD Open-Source S.I. Botched, Hope For The Future

    We're now going into eight months since the AMD Radeon HD 7000 series "Southern Islands" graphics cards first launched. In that time the Catalyst Linux support has been stable and fine, but the open-source driver support is still unusable...

    http://www.phoronix.com/vr.php?view=MTE0OTY

  • #2
    I don't understand why AMD always does their best to look like idiots. This is what makes the difference between them and INTEL or NVIDIA.
    Last edited by bulletxt; 07-30-2012, 08:31 AM.

    Comment


    • #3
      Originally posted by bulletxt View Post
      I don't understand why AMD always does their best to look like idiots. This is what makes the difference between them and INTEL or NVIDIA.
      Yeah, right. NV is completely ignoring open source GPU support. Non existing drivers avoid news about some missing or lackluster features.
      Great logic you applied there.

      Comment


      • #4
        Originally posted by dhewg View Post
        Yeah, right. NV is completely ignoring open source GPU support. Non existing drivers avoid news about some missing or lackluster features.
        Great logic you applied there.
        I did not say "A" is better than "B". I said "A" generally tends to behave in a way that they look idiots and not professional when compared against "B" or "C". Look at it from an economical/business point of view. AMD brand looses points when they do such shit. And do you know how important is brand? Think about it, leave the OSS vs blob fight for the geeks.

        Comment


        • #5
          Originally posted by bulletxt View Post
          I did not say "A" is better than "B". I said "A" generally tends to behave in a way that they look idiots and not professional when compared against "B" or "C". Look at it from an economical/business point of view. AMD brand looses points when they do such shit. And do you know how important is brand? Think about it, leave the OSS vs blob fight for the geeks.
          Eh? Neither did I claim that you said A is better than B. My argument about your logic still stands.
          Because bridgman posts here and admits that things didn't work out as planned makes them idiots? I'd call that honest.

          Since that article is about the open source radeon stack, your "OSS vs blob" statement makes even less sense.

          Comment


          • #6
            It's not that AMD is doing a bad job. It's just that Michael seems to like the pageviews he's getting from negative headlines.

            The OS drivers are not "botched", they're "right on track, but not finished yet". Previous generations had to wait half a year or longer, too. While that delay has been shortened with each successive generation, it's not surprising that the architectural HW changes would cause additional delays this time.

            In fact, AMD has been very clear about this upfront. They said that they started on OSS work earlier, but that the headstart would likely be eaten up by the additional work required. They said so before HD7xxx-hardware could be bought.

            So if any customer is unsatisfied that OSS support is taking this long, it's the customer's fault for expecting a feature earlier than it was promised. AMD has been very honest with that they will do, what they try to do and what they can't do, and all they're getting for it is repeated rants that those "will do"s don't include a pony.

            The worst part is that Michael has to know all this. We know he's reading all of bridgman's posts because he likes to copy/paste them for free articles like this one. Yet he's still continuing to publish articles that label the current state of the OSS drivers as a failure on AMD's part, despite the project being roughly inside the initial estimates.

            Comment


            • #7
              Originally posted by phoronix View Post
              In that time the Catalyst Linux support has been stable and fine
              I LOL'ed so hard when I read this.

              Comment


              • #8
                Not many community developers are interested in hacking on Radeon GPU power management code, unfortunately.
                So hire them. If you pay them I'm sure they will have more interest. More easy to saying then doing, but that's the point in the end.

                Comment


                • #9
                  Originally posted by RealNC View Post
                  I LOL'ed so hard when I read this.
                  Yeah?

                  Originally posted by phoronix
                  In that time the Catalyst Linux support has been stable and fine
                  Meanwhile on my Notebook (with 7970M) :

                  Originally posted by fglrx
                  No supported adapters detected
                  I don't find that funny

                  Comment


                  • #10
                    It's funny... I was expecting an news post more like "Open source SI fixes pushed, glxgears now runs, open source OpenCL work proceeding on schedule". Oh well...
                    Last edited by bridgman; 07-30-2012, 11:19 AM.

                    Comment


                    • #11
                      hmm I thought, yes it kind of sucks but at least a person who talks like a human about process.

                      No press bullshit where all is good and for anything goes wrong others are to blame...

                      Comment


                      • #12
                        These endless SI "stories" from Michael are getting tiresome. The are little more then just restating "SI is unusable" with a a few other negative comments. The drivers are working just fine for display purposes and performance is fine for non-3D tasks. Also, we never gave any explicit time frames for when any specific features would be "finished." Display support was out around launch time for SI so the hw has been "usable" since then. From the beginning we've stated and re-stated that the open source project is designed to support the open source community, not write the entire stack ourselves and that the rate of progress will largely be defined by the amount of community involvement. We've put a lot of effort in to jump start development and get usable code and support out there to help the community. Unfortunately very few developers outside of AMD have done much with SI. At this point it's mostly just AMD developers working on it. Unfortunately our time is shared with other projects that many would consider equally important: improving 2D support (glamor), adding OpenCL support, adding support for new asics, and other projects. I don't see how SI support was "botched." We did exactly what we said we would do. Secondly for things outside of display and 3D (like UVD), we said from the beginning and have repeated on many occasions (included quoted references in the article), that we aren't committing to supporting them at the moment, but we will look into whether we can support them in the future and still meet appropriate contractual obligations. That may mean we can, that may mean we can't.

                        Comment


                        • #13
                          Originally posted by agd5f View Post
                          These endless SI "stories" from Michael are getting tiresome. The are little more then just restating "SI is unusable" with a a few other negative comments. The drivers are working just fine for display purposes and performance is fine for non-3D tasks. Also, we never gave any explicit time frames for when any specific features would be "finished." Display support was out around launch time for SI so the hw has been "usable" since then. From the beginning we've stated and re-stated that the open source project is designed to support the open source community, not write the entire stack ourselves and that the rate of progress will largely be defined by the amount of community involvement. We've put a lot of effort in to jump start development and get usable code and support out there to help the community. Unfortunately very few developers outside of AMD have done much with SI. At this point it's mostly just AMD developers working on it. Unfortunately our time is shared with other projects that many would consider equally important: improving 2D support (glamor), adding OpenCL support, adding support for new asics, and other projects. I don't see how SI support was "botched." We did exactly what we said we would do. Secondly for things outside of display and 3D (like UVD), we said from the beginning and have repeated on many occasions (included quoted references in the article), that we aren't committing to supporting them at the moment, but we will look into whether we can support them in the future and still meet appropriate contractual obligations. That may mean we can, that may mean we can't.
                          well accounting the massive job you are doing here rewriting and updating the entire linux graphic stack [and as happy r600g user] i give you my kudos and in many areas r600g has proven to be massively superior than fglrx[2d/XV for example even though exa is getting old(itching to test glamor but im with xorg 1.13 so im waiting the patch)] but is true that r600g has some rough edges and mesa need more work but is true also that the situation is improving daily so no complains here

                          is understandable too that GCN been an massive change from VLIW will require lots of work and testing hence stretching the release time but that is expected in any other hardware vendor.

                          maybe an advice is tag OSS drivers as "alpha state under heavily delevopment" or something similar and get newbies to stick with the blob for now so you can avoid the typical LOL questions like UVD .... but well haters are gona hate but be sure that most of us trully respect and appreciate your work and AMD involvement in the OSS drivers [i do understand the fact AMD is not here to write the driver but to provide documentation and help to the community, dunno why people seem to have such a hard time understanding that tho] and well i praise Intel and nouveau group too [not Nvidia tho (.!. linus style)]

                          as side note i tried some time ago to help with the vdpau tracker and i must say gallium code is kinda creepy [at least for a c++ dev ] so maybe some documentation could help to bring more guns[i know is hard tho so is like an idea to keep around no pressure], either way just ignore the haters and keep your awesome job

                          Comment


                          • #14
                            Originally posted by agd5f View Post
                            These endless SI "stories" from Michael are getting tiresome. The are little more then just restating "SI is unusable" with a a few other negative comments. The drivers are working just fine for display purposes and performance is fine for non-3D tasks.
                            Michael is provoking imho, and be aware of his target: the linux user. From this point of view the radeon driver is not good, at least when you talk about SI. Put in other words, for the normal PC user, one who can't distinguish a fan from a CPU, radeon is not usable I think. The user would have a sluggish experience on SI I think, and you can't give him a reason but "the driver is not ready". I was a radeon user, and it was a bit (not a lot, but still) of a pain I have to say honestly. I respect and understand your work Alex and I can see you are tired by article like this. But I really think Micheal is just warning the average user (and hopefully his laud will gain some attention which is important to get new developers), for his our benefit of course, this site is his work like radeon is part of yours.

                            And on my personal POV: this policy of AMD of just supporting the open-source community.... well I don't like it very much, your team is too small. Community driven drivers rarely success, expecially if complicated as GC drivers. The manufacter must drive the development, not just support it, but you need more HR for this indeed. You can't do that in just 4.

                            Anyway I wish you best luck with radeon 8000 series.... who knows you might surprise me and I will start buying AMD again.
                            Cheers

                            Comment


                            • #15
                              Originally posted by enrico.tagliavini View Post
                              Michael is provoking imho, and be aware of his target: the linux user. From this point of view the radeon driver is not good, at least when you talk about SI. Put in other words, for the normal PC user, one who can't distinguish a fan from a CPU, radeon is not usable I think.
                              I would posit just the opposite. In that case, the user probably can't tell whether he's using vesa or a native driver. Both will put an image on the screen a perform quite well for non-3D tasks which is what the vast majority of users do. Even for basic 3D, llvmpipe should do a pretty decent job in the interim. For that, the driver already works quite well.

                              Comment

                              Working...
                              X