Announcement

Collapse
No announcement yet.

"Ask ATI" dev thread

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

  • Your response to my last post, Bridgeman, was actually quite enlightening and I have a much better idea of how the process works. As for the new multi-head support, I didn't even notice or know about it until after I made my post. When I saw that, I was suddenly very happy about what this new driver brought to the table. I frequently use my desktop, and now my laptop, to output movies and such to my tv over the HDMI connection, so this is a feature that I will make heavy use of. Well... once the issue with audio over hdmi is corrected haha. But, if that is the worst of my problems with fglrx, then you guys are doing a good job

    Comment


    • Speaking of power management, I've tuned every knob possible on my new HP dv5z with HD3450 but it still runs hotter and with shorter battery life under Linux than under Windows. Since I see the CPU is idling at it's lowest power state (500MHz) I'm guessing the GPU is to blame here, and the lack of a PowerPlay-like feature on the Linux side. I would really like to be able to underclock/undervolt this GPU when the notebook is running on batteries.

      Comment


      • Update on checkerboard bug

        Having installed the Catalyst 8.8 driver (doing a proper clean up before), I can now report that the mplayer -vo gl2 fullscreen/windowed switch doesn't trigger the 'checkerboard of doom' bug anymore (not even a glitch). However:
        - the glitch is still present when Nexuiz 2.4.1 starts
        - Wine still goes FUBAR (give me back my WoW!)
        - the pattern is however less scrambled and is becoming 'predictable'.

        Another bug, which must be found in the shaders code, is when using Celestia 1.5 with the OpenGL 2.0 render path: it will start by working well, but then suddenly shaded objects will get a quilted pattern of red, green, yellow, blue etc. tint. This will stay on until the driver is restarted (and X becomes unable to close normally, having to revert to a hard X Ctrl-Alt-Backspace exit ain't no idea of fun).

        Last, it would be nice if texture compression could be deactivated one way or another (environment variable, control panel setting, etc.): that would help when using mplayer -vo gl2, as it currently looks washed out - but is also the only way to get vsync.

        Comment


        • With the OpenGL 3.0 spec out, is that a priority now for any development? Actually any thoughts about 3.0 and ati's drivers would be interesting.

          Comment


          • Originally posted by bridgman View Post
            One of the problems here is that a lot of the things you think of as "bugs" (eg video tearing) are actually "lack of features". Syncing video playback to vblank is one of the areas where the current X/DRI stack has always had problems (and is in the process of being re-written again), while Windows and MacOS have had a framework to support this for years.
            Heh, I wonder how long it will take to get past the legacy of the XFree86 team. Ah well, enlightening and thanks for explainig why it's not "fixed".


            I'm actually surprised that nobody is talking about the improved multi-head support we added in 8.8. That is by far the most disruptive change but it is also one of the most requested features we hear about.
            Not by me. I'd rather go to a 30" screen than 2x24". My real need is for something to play nice with mythtv with a single output device.


            We do understand your needs (and your frustration) and that was one of the reasons we kicked off the open source driver initiative last year. I expect that over time most consumer users will gravitate to the open source driver, since it is targeted precisely at those requirements -- simple driver, core functionality, and able to be tweaked to match your specific distribution so things like suspend/resume work reliably for relatively more users.

            ...

            The proprietary driver will be the vehicle for delivering maximum performance and functionality, but I see it as an "upgrade path" for consumer users if they are looking for something more than the open source driver can provide.
            Fair enough and we (I) appreciate the genuinely excellent steps in the right direction. What I didn't understand until recently is that the very well received 780G chipset won't work for me. I need a small quiet HTPC frontend. Something that will work with Win32 powerdvd and mythtv. For now and for the forseeable future, I can't have a HD3200 and mythtv without tearing. That's not acceptable. My temporary workaround was to throw an NVidia card into my HTPC. Mythtv now looks great and Powerdvd still works. But I've blown my heat budget and now the FE can't be both quiet and stable ...

            ... so I've ordered an intel G45 mobo & E8500 CPU ...

            we'll see how mythdora 5 with new kernel, X.org & Mesa (or how well fedora 10 alpha does). Maybe I'll get my requirements met.

            This isn't what I wanted and I'm sure AMD would rather not have other customers coming to the same conclusion.

            A (likely to be hit) target date for fixing the tearing and checkerboarding problem [from ANY driver!!!] would help planning.

            PS - In time I'll require p24 output on both platforms and Intel may have a problem with that (they currently do).

            Thanks for the forum participation

            Comment


            • Originally posted by bridgman View Post

              I see some posts about "why are you adding features like overclocking ?" but overdrive, powerplay and thermal management are all different aspects of the same code. As we finish integrating power management into the Linux drivers things like Overdrive more-or-less come along for free. I guess we could leave them completely hidden but I don't think that would save much time or eliminate any bugs because we need to set everything up properly anyways.
              Ok, I discovered some weeks ago that through the pplib-cmd trick I can get a good control/monitoring ability on my HW, but now comes a question: why being on a shell without Xorg, the clock of my Mobility Radeon jumps to the highest level??
              and why if I limit it through the --pplib-cmd "notify psrc dc", the shell starts to be difficulty responsive? (with that lower clock, I often need to press a letter 2 times to see it typed on the shell)
              (I know it's the same point, simply 2 different aspects)


              Originally posted by bridgman View Post
              I'm actually surprised that nobody is talking about the improved multi-head support we added in 8.8. That is by far the most disruptive change but it is also one of the most requested features we hear about.
              Because it doesn't work!!!
              I noticed the wonderful new GUI, I liked it, but Dual-Head is still broken!

              The big desktop creates a corruption on the new screen, and if I restart Xorg with the new monitor plugged in, first Xorg is not able to start, and then, after a second login, the computer reboots

              Nice, ain't it?

              Originally posted by bridgman View Post
              We do understand your needs (and your frustration) and that was one of the reasons we kicked off the open source driver initiative last year. I expect that over time most consumer users will gravitate to the open source driver, since it is targeted precisely at those requirements -- simple driver, core functionality, and able to be tweaked to match your specific distribution so things like suspend/resume work reliably for relatively more users.

              Right now there are two remaining things which need to be addressed in order to make the open source driver viable for general consumer use -- addition of a decent memory manager in the kernel (which allows the implementation of GL 2.0 functionality), and support for the 6xx/7xx 3d engine (which allows more acceleration to be added to the HD2xxx-HD4xxx drivers than the shadowfb acceleration we have today). The first task is being worked on by the community, with Dave Airlie taking the lead, and the second task is being worked on by AMD and Novell developers.

              The proprietary driver will be the vehicle for delivering maximum performance and functionality, but I see it as an "upgrade path" for consumer users if they are looking for something more than the open source driver can provide.
              The problem for the open source driver right now, is that it has no power management on my HW (Mobility Radeon HD2600).

              Comment


              • @Bridgman:

                I recently put together a HTPC MythTV box with AMD 780G chipset and integrated ATI Radeon HD 3200 graphics and ran into unexpected problems.

                1. Can you provide an estimated timeframe when the video tearing/shearing problem related to the lack of vsync will be fixed?

                2. The Catalyst 8.8 drivers are very unstable for me. I usually get uptimes anywhere from 5 minutes to a max of 6 hours before the system crashes and reboots. The HTPC box has not crashed at all with the Catalyst 8.4 drivers. Any insight on this?

                For the MythTV HTPC setup, I specifically chose the motherboard for the ATI 780G/Radeon HD 3200 chipset used and the advertised Linux compatibility. The unexpected video shearing/tearing issue is a disappointment and I'm looking at my options.

                Dan
                Last edited by rtpbiker; 09-07-2008, 06:26 PM.

                Comment


                • X.Org 1.5? If you ever tried it, you can't live without 1.5's new mouse "predictive" acceleration thingy anymore.

                  Comment


                  • Well, if Keith is able to make good on his call for Xorg 1.6 before the year's end, it may not be something to focus on just yet. It doesn't make much sense to code for compat for 1.5, when they can do both at the same time around when 1.6 is released.

                    Comment


                    • I've only just started using two monitors, but I can say that using aticonfig worked a treat for basic dual head setup. I say basic because while xinerama 2d works just fine, 3d apps won't start (just a black screen / window until I kill the process) but won't kill off X either. Non-xinerama has no problems at all.
                      Never been able to open amdcccle. Always some error about Xlib: sequence lost (blah blah blah). So I can't comment on using that with dual head stuff.
                      This is all using enlightenment (e16), with gentoo (64bit) and a radeon1600. I usually refrain from posting comments as I generally stick with unstable packages.

                      Comment


                      • Originally posted by mirv View Post
                        Never been able to open amdcccle. Always some error about Xlib: sequence lost (blah blah blah). So I can't comment on using that with dual head stuff.
                        This is all using enlightenment (e16), with gentoo (64bit) and a radeon1600. I usually refrain from posting comments as I generally stick with unstable packages.
                        Uhm, have you switched to ati opengl? Just make sure to run "eselect opengl ati". This sounds like you maybe have forgotten to do so.

                        Comment


                        • Originally posted by ivanovic View Post
                          Uhm, have you switched to ati opengl? Just make sure to run "eselect opengl ati". This sounds like you maybe have forgotten to do so.
                          I usually run that (actually "eselect opengl set ati"), although the gentoo install scripts do it anyway.
                          Full error message is:
                          Code:
                          Xlib: sequence lost (0x10000 > 0xb) in reply type 0x0!
                          X Error: 0 0
                            Major opcode: 0 ()
                            Resource id:  0x0
                          It's just amdcccle, nothing else does it. Does it still use qt? Tried updating that too, no avail.

                          Comment


                          • that eselect stuff is nice - but sometimes it fails to set the symlinks correctly (when there is some symlink already it does not expect and other little things). So have you checked with glxinfo that everything is ok?

                            Comment


                            • glxinfo is fine, glxgears runs fine, my "test bed" of games (ut2004, doom3,ppracer) all work witout a glitch, my own opengl development runs fine (when I've coded it properly!). Nothing in dmesg, but after work I can run through the full xorg log. amdcccle is the only program that doesn't run. I'm not all that fussed about it personally - it's more that I want it to run so that I can help others.

                              -- Nothing out of the ordinary from the x log or any kernel logs. User privileges are fine.
                              Last edited by mirv; 09-08-2008, 12:19 PM.

                              Comment


                              • Originally posted by wfeltmate View Post
                                Well, if Keith is able to make good on his call for Xorg 1.6 before the year's end, it may not be something to focus on just yet. It doesn't make much sense to code for compat for 1.5, when they can do both at the same time around when 1.6 is released.
                                Why doesn't it make sense? I'm not a fan of last-minute implementations. Do it early, and you'll discover potential problems early. Being ready for prime time before the actual arrival of prime time is the smart thing to do.

                                Comment

                                Working...
                                X