[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
Don't think this is a bug. There is definitely a performance regression using the intel driver on version 2.5.1 in jaunty. Bug # is 252094. Link: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video- intel/+bug/252094 -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
Conn, Executing glxgears with the vblank_mode variable set as you described makes no performance difference on my Acer Aspire One notebook. Likewise, disabling vertical sync using the driconf utility does not offer any performance gains in glxgears. Can you offer any further explanation or instruction? -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
Pandemonium, I was addressing citizenofnowhere, and his issue is almost certainly due to vblank_mode being enabled - but he is using new version of the intel and libdrm code. I don't have access to any Intel hardware in order to test at the moment, so I can't comment any further. If you see a glxgears result that's constrained to your refresh rate (e.g., 60-62fps for a laptop, or a rate corresponding to your display device such as 70, 75, 85), then the problem is vblank_mode (and perhaps the environment variable is broken/ignored in the new driver, so it cannot be overridden). I can't say if the problem applies to you. From my understanding of events, we're kind of in a transitional stage with regards to DRI. We had the introduction of DRI2 which was initially supposed to be in the form of TTM, which then got dropped in favour of GEM, and I've been reading some complaints that GEM is too Intel- centric, thus unsuitable for other chipsets. I have no idea what the situation is at the moment, but I honestly don't think the Xorg developers are concerned so much with performance for the legacy DRI code at this time, which is unfortunate for users (but also understandable). Bryce knows a lot more, and he's already explained the situation (re: performance of the 2.4 vs 2.5 driver and absence of the legacy INTEL_BATCH variable). -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
Too bad, as the performance regression is quite notable also on a aspire one. -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
I'll give this a try. I think I'll wait for the final release of 8.10, and base my attempts off of that. Hans -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
I gave it a try, but the build instructions mention an autogen script that is missing from the driver source package. I tried ./configure, and it ran, telling me I was missing some files that I don't know how to install. If I find some more time, I'll work on this some more. Hans -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
Hans I'll try it out on my AAO and if its any better i'll package it -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
It got worse. I compiled the 2.5 intel driver and recompiled libdrm2 to version 2.4.1 - FPS dropped from 350 (about averagish without intel_batch) to 75. (this is with no compositing on, with compositing it's like 40!) Driver section in Xorg is clean, I let X autodetect everything. I've tried the vblank_mode workarounds from https://bugs.freedesktop.org/show_bug.cgi?id=18128 with the newly compiled driver. If you want I can provide logs and stuff. This is a pretty serious regression IMO. -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
That is not a regression, it's a configuration change that won't adversely affect performance like you seem to imagine. Your glxgears performance have been constrained to the refresh rate of your display device (75Hz, judging from your results). This is perfectly normal behaviour. Even a top of the line NVIDIA card using the non-free drivers will constrain to the refresh rate when VSync (the equivalent of vblank_mode) is active. Nothing to see here, move along... -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
Re: [Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
Ahem, What you are saying about VSync sounds right. Do you have any suggestions for annafil to help get a more realistic glxgears score? Hans Conn wrote: That is not a regression, it's a configuration change that won't adversely affect performance like you seem to imagine. Your glxgears performance have been constrained to the refresh rate of your display device (75Hz, judging from your results). This is perfectly normal behaviour. Even a top of the line NVIDIA card using the non-free drivers will constrain to the refresh rate when VSync (the equivalent of vblank_mode) is active. Nothing to see here, move along... -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
Hans, the environment variable vblank_mode controls this behaviour. You can revert to the original behaviour for glxgears like so: $ vblank_mode=0 glxgears There is also a configuration utility in our repositories called driconf, which can expose and set all the DRI configuration options, including vblank_mode. -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
I spoke with upstream about this, and confirmed that the option is gone. Regarding the performance regression, they're not so interested in performance reports for 2.3 - 2.4. They'd be interested in hearing about performance issues from 2.3 to 2.5, however we don't have a test package for that at this time. If you do get a chance to test 2.5 and find there to still be a measurable performance regression, please file a new bug and let me know. ** Changed in: xserver-xorg-video-intel (Ubuntu) Status: Triaged = Won't Fix -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
Yeah, that was an experimental debug parameter for the old version of -intel, and upstream removed it (and the code it influenced) from the 2.4 driver we carry in Intrepid. Since you've done some measures and included evidence from hardy and intrepid, then this bug may be worth forwarding upstream for their comment. However, I'm pretty sure INTEL_BATCH is gone, obsolete, and no longer supported. But perhaps the performance difference could be looked at closer. ** Changed in: xserver-xorg-video-intel (Ubuntu) Importance: Undecided = Medium Status: Incomplete = Triaged -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
** Attachment added: xorg.conf_hardy.txt http://launchpadlibrarian.net/18683213/xorg.conf_hardy.txt -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
** Attachment added: Xorg.0.log_hardy.txt http://launchpadlibrarian.net/18683220/Xorg.0.log_hardy.txt -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
** Attachment added: xorg.conf_intrepid.txt http://launchpadlibrarian.net/18683223/xorg.conf_intrepid.txt -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
** Attachment added: Xorg.0.log_intrepid.txt http://launchpadlibrarian.net/18683230/Xorg.0.log_intrepid.txt -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
** Attachment added: lspci-vvnn_intrepid.txt http://launchpadlibrarian.net/18683256/lspci-vvnn_intrepid.txt -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
** Attachment added: lspci-vvnn_hardy.txt http://launchpadlibrarian.net/18683248/lspci-vvnn_hardy.txt -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 284082] Re: INTEL_BATCH=1 not implemented in Intrepid ?
Hi coolos, Please attach the output of `lspci -vvnn`, and attach your /var/log/Xorg.0.log file from after reproducing this issue. If you've made any customizations to your /etc/X11/xorg.conf please attach that as well. ** Changed in: xserver-xorg-video-intel (Ubuntu) Status: New = Incomplete -- INTEL_BATCH=1 not implemented in Intrepid ? https://bugs.launchpad.net/bugs/284082 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
