Announcement

Collapse
No announcement yet.

Likely Radeon Gallium3D Regression On Linux 3.14 + Mesa 10.2

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

  • #11
    Originally posted by genstorm View Post
    When they matter, e.g. in final versions, yes. Have YOU ever used software based on live sources? Regressions may happen on any commit, might be fixed in one of the following commits. In that case, where HyperZ has been disabled, you will get a fine Changelog entry with the final announcement, and all of those costly benchmarking and wondering 'what the heck degraded here' moments have been a waste of time. Time that is so badly missing to improve the poor average quality of other articles.

    You can keep your insults to yourself.
    I do.

    I'm literally using Mesa git.
    And I recommend it to anyone who use OpenSuSE (pontostroy repos) or Ubuntu (oibaf repos).
    Its easy, Its quick, and very benefitial.


    And for sure its good to know that form now I need to set that HyperZ variable.

    ("But its beta" is good excuse for CLOSED source projects, where onlything You can do is hope for the best, and only priorities of devs matter. In FLOSS You can grapple with the code Yourself. As some comment above showed.)
    (Never played AB, though I have nothing against them nor against casual gamers, You included. As long as You do not demand that others start changing their behavior.)
    Last edited by przemoli; 03-02-2014, 07:33 AM.

    Comment


    • #12
      Originally posted by przemoli View Post
      I'm literally using Mesa git.
      And I recommend it to anyone who use OpenSuSE (pontostroy repos) or Ubuntu (oibaf repos).
      Its easy, Its quick, and very benefitial.
      You should never ever recommend it to any _regular_ user.

      Originally posted by przemoli View Post
      And for sure its good to know that form now I need to set that HyperZ variable.
      Something that people found out by simply looking through the last few commits between 'normal' and 'degraded' state.

      Originally posted by przemoli View Post
      ("But its beta" is good excuse for CLOSED source projects, where onlything You can do is hope for the best, and only priorities of devs matter. In FLOSS You can grapple with the code Yourself. As some comment above showed.)
      It doesn't matter for the majority of users to whom software development is a black box. They will not look into gitweb and search for the regression, but well, they shouldn't use it in the first place. They will be bothered by the news item though, while it is entirely possible that it will not matter for a final release and if, is going to be properly announced (or even taken care of by package maintainers).

      Comment


      • #13
        Hm I never seem to get any problem with hyperz so how would I enable it? I using the oibaf PPA.

        Change the flag to DBG_HYPERZ and reverse the logic
        so setting the flag enabled the feature. This disables
        hyperz on r600g and radeonsi by default. It can be
        enabled by setting the env var.

        Where can i apply the settings please?

        Thank you

        Comment


        • #14
          Originally posted by genstorm View Post
          When they matter, e.g. in final versions, yes. Have YOU ever used software based on live sources?
          I think mostly everybody who actually play games with open source drivers use Oibaf PPA or other sources with fresh Mesa as well as latest kernel. Also as games I mean not only SuperTuxKart or OpenTTD.

          PS: I use those drivers from git for about a year and never get any serious glitch or GPU lockup because of that.
          Last edited by _SXX_; 03-02-2014, 07:52 AM.

          Comment


          • #15
            Originally posted by jonty789 View Post
            Hm I never seem to get any problem with hyperz so how would I enable it? I using the oibaf PPA.
            Set this environment variable:
            R600_DEBUG=hyperz

            Comment


            • #16
              Originally posted by jonty789 View Post
              Hm I never seem to get any problem with hyperz so how would I enable it? I using the oibaf PPA.

              Change the flag to DBG_HYPERZ and reverse the logic
              so setting the flag enabled the feature. This disables
              hyperz on r600g and radeonsi by default. It can be
              enabled by setting the env var.

              Where can i apply the settings please?

              Thank you
              Code:
              env R600_DEBUG=hyperz your_app
              edit.
              too late

              Comment


              • #17
                Originally posted by _SXX_ View Post
                I think mostly everybody who actually play games with open source drivers use Oibaf PPA or other sources with fresh Mesa as well as latest kernel. Also as games I mean not only SuperTuxKart or OpenTTD.

                PS: I use those drivers from git for about a year and never get any serious glitch or GPU lockup because of that.
                Then, lucky you. I understand that people using AMD open source drivers have for a long time been used to running on the edge of only the latest versions of software, even from git, because of the huge backlog they had to clear until recently. For the several AMD boxen I happen to take care of, I did that myself, emerged recent versions, tested them, then handed over the system to its users if OK. However, between a non-bootable system, black screens, no X or simple graphics corruption, anything can happen. That's why, given the choice, you simply don't advertise this to regular users. And you are wrong if you think the majority is doing it.

                Comment


                • #18
                  Originally posted by genstorm View Post
                  Then, lucky you. I understand that people using AMD open source drivers have for a long time been used to running on the edge of only the latest versions of software, even from git, because of the huge backlog they had to clear until recently. For the several AMD boxen I happen to take care of, I did that myself, emerged recent versions, tested them, then handed over the system to its users if OK. However, between a non-bootable system, black screens, no X or simple graphics corruption, anything can happen. That's why, given the choice, you simply don't advertise this to regular users. And you are wrong if you think the majority is doing it.
                  For that matter I anti-recommend Catalyst.

                  Its more likely to get You into troubles. (And well. Clearing bad Mesa/kernel duo is as simple as using legacy kernel from Advanced options at boot time. Try that for Catalyst...)

                  Comment


                  • #19
                    People have to report the regression first!

                    Originally posted by genstorm View Post
                    When they matter, e.g. in final versions, yes. Have YOU ever used software based on live sources? Regressions may happen on any commit, might be fixed in one of the following commits.
                    The sooner a regression is reported, the sooner it can be fixed. Why *on Earth* should anyone wait for a "final" version to be released first, when that *by definition* is going to impact ordinary users?

                    Thanks for noticing this, Michael.

                    Comment


                    • #20
                      I beg to differ:

                      Originally posted by chrisr View Post
                      The sooner a regression is reported, the sooner it can be fixed.
                      That's when you report it upstream as soon as you notice.

                      Originally posted by chrisr View Post
                      Why *on Earth* should anyone wait for a "final" version to be released first
                      That's when you write an article about it.

                      Remember when Michael bisected a regression in the kernel, then decided to _not_ disclose the bad commit until several days later?

                      Comment

                      Working...
                      X