Bug 656060

Summary: intel [945G] Very poor compiz performance
Product: [openSUSE] openSUSE 11.4 Reporter: Atri Bhattacharya <badshah400>
Component: X.OrgAssignee: Stefan Dirsch <sndirsch>
Status: RESOLVED DUPLICATE QA Contact: E-mail List <xorg-maintainer-bugs>
Severity: Major    
Priority: P3 - Medium CC: dimstar, Martin.Seidler
Version: Milestone 5 of 6   
Target Milestone: ---   
Hardware: x86-64   
OS: SUSE Other   
Whiteboard:
Found By: --- Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Attachments: Screenshot while moving the YaST control centre window around

Description Atri Bhattacharya 2010-11-26 10:13:19 UTC
Created attachment 402261 [details]
Screenshot while moving the YaST control centre window around

User-Agent:       Mozilla/5.0 (X11; Linux x86_64; rv:2.0b7) Gecko/20101105 Firefox/4.0b7

Running GNOME from Live-CD, Milestone 3.

The performance when using compiz is very poor. Everything in motion (e.g. moving windows, rotating desktop cube, etc.) causes heavy flickering of the window/screen. Refresh rates are also extremely poor making animation effects look choppy and dysfunctional. The attached screenshot taken while moving the YaST control centre around shows this to an extent.

The following is the configuration for my graphics chipset

16: PCI 02.0: 0300 VGA compatible controller (VGA)
  [Created at pci.318]
  Unique ID: _Znp.U+waTKLi8tD
  SysFS ID: /devices/pci0000:00/0000:00:02.0
  SysFS BusID: 0000:00:02.0
  Hardware Class: graphics card
  Model: "Intel 945G"
  Vendor: pci 0x8086 "Intel Corporation"
  Device: pci 0x2772 "945G"
  SubVendor: pci 0x103c "Hewlett-Packard Company"
  SubDevice: pci 0x3022 
  Revision: 0x02
  Driver: "i915"
  Driver Modules: "drm"
  Memory Range: 0xe2000000-0xe207ffff (rw,non-prefetchable)
  I/O Ports: 0xc000-0xc007 (rw)
  Memory Range: 0xd0000000-0xdfffffff (ro,non-prefetchable)
  Memory Range: 0xe2080000-0xe20bffff (rw,non-prefetchable)
  IRQ: 16 (281448 events)
  I/O Ports: 0x3c0-0x3df (rw)
  Module Alias: "pci:v00008086d00002772sv0000103Csd00003022bc03sc00i00"
  Driver Info #0:
    XFree86 v4 Server Module: intel
  Driver Info #1:
    XFree86 v4 Server Module: intel
    3D Support: yes
    Extensions: dri
  Config Status: cfg=no, avail=yes, need=no, active=unknown


Reproducible: Always

Steps to Reproduce:
1.Enable desktop effects
2.Make windows move, or rotate desktop cube
Comment 1 Dominique Leuenberger 2010-11-26 10:20:52 UTC
did this work properly in openSUSE 11.3? If so, then this would be a regression, most likely in the graphic driver.

If this did not work on older releases neither, then I'd be inclined to blacklist this chip for compiz.
Comment 2 Atri Bhattacharya 2010-11-26 10:50:47 UTC
(In reply to comment #1)
> did this work properly in openSUSE 11.3? 

Flawlessly :)
Comment 3 Dominique Leuenberger 2010-11-26 11:05:59 UTC
Assigning to the X.org team.

This sounds like a regression in the intel driver, which is exposed by compiz.
Comment 4 Martin Seidler 2010-11-29 20:03:18 UTC
I get the same special effect if I change to an other workspace. But not with moving a window - I just cannot move any window with desktop effects on.

With
openSUSE 11.4 Factory 2.6.37-rc3-git1-6 GNOME 2.32.1
Pentium M 740 Intel 915GM 1400x1050

(But before that sliced windows when moving
- since Milestone 3 and since I was able to turn the desktop effects on at all).

And by the way - every time I tuned desktop effects off - they were tuned on again after I booted.
Comment 5 Martin Seidler 2010-12-16 12:45:07 UTC
(In reply to comment #4)
openSUSE 11.4 factory-tested (now saying it is "Milestone 5" - until Milestone 3 updated to factory or factory-tested)
Linux 2.6.37-rc5-12-desktop GNOME 2.32.1 i686 i386

The windows are moving again (not only until the update form today) but the sliced moving windows effect stays on.

> And by the way - every time I tuned desktop effects off - they were tuned on
> again after I booted.
I opened an additional bugreport for that:

Bug 659886
Desktop Effects re-enable themselves after log out of GNOME/after shutdown with Compiz ( via Simple CCSM ) on openSUSE 11.4
https://bugzilla.novell.com/show_bug.cgi?id=659886
Comment 6 Martin Seidler 2010-12-16 12:49:11 UTC
Should I update to "Version: Milestone 5 of 6"?
Could someone confirm that on a x64-64 system for the latest version of factory/factory-tested/Milestone? Or should I change that to i686 ?

Regards
Martin
(pistazienfresser)
Comment 7 Martin Seidler 2010-12-18 01:18:08 UTC
Compare also the resolved bug in openSUSE 11.3
(also problems with Compiz 3-D desktop effects with several Intel graphics):

Bug 597596 - Compiz breaks gnome desktop in oS 11.3 M5, M6
https://bugzilla.novell.com/show_bug.cgi?id=597596
Comment 8 Martin Seidler 2010-12-20 09:30:27 UTC
Bug 655937 - Regression in kernel 2.6.37 with regards to i915 driver (Intel Graphics)
https://bugzilla.novell.com/show_bug.cgi?id=656060
Opened: 2010-11-25 09:42 UTC

with

"
[...]
Based on the kernel-desktop I build the rc4-git4 kernel and this resolved the
described issue. Transparency and Wobbly Windows are working normal again. 

It might be good to build this version of the kernel and to get it into
Factory.
[...]
"

the same bug?
Comment 9 Stefan Dirsch 2010-12-20 10:08:26 UTC
Sounds reasonable.

*** This bug has been marked as a duplicate of bug 655937 ***
Comment 10 Martin Seidler 2010-12-20 11:33:20 UTC
(In reply to comment #9)
> Sounds reasonable.
> 
> *** This bug has been marked as a duplicate of bug 655937 ***

But I think at least one of these report threads should stay open ;)

And if it should be the other one (bug 655937), maybe someone could edit the list for the
"Most annoying bugs" since M4, too?

See:
http://en.opensuse.org/openSUSE:Most_annoying_bugs_11.4_dev#openSUSE_11.4_Milestone_4_of_6

-> 

Regards
Martin
(pistazienfresser)
Comment 11 Martin Seidler 2010-12-20 11:37:02 UTC
(In reply to comment #10)
> (In reply to comment #9)
> > Sounds reasonable.
> > 
> > *** This bug has been marked as a duplicate of bug 655937 ***
> 
> But I think at least one of these report threads should stay open ;)
> 
> And if it should be the other one (bug 655937), maybe someone could edit the
> list for the
> "Most annoying bugs" since M4, too?
> 
> See:
> http://en.opensuse.org/openSUSE:Most_annoying_bugs_11.4_dev#openSUSE_11.4_Milestone_4_of_6
> 

Sorry, the other one *is* open, but maybe it should have at least the same priority as this report had?
Comment 12 Stefan Dirsch 2010-12-20 13:02:40 UTC
> maybe someone could edit the list for the "Most annoying bugs" since M4, too?
> 
> See:http://en.opensuse.org/openSUSE:Most_annoying_bugs_11.4_dev#openSUSE_11.4_Milestone_4_of_6

It's a Wiki used and written by our community. Could you take care of that as part of our community ? Thanks a lot!
Comment 13 Stefan Dirsch 2010-12-20 13:04:50 UTC
> Sorry, the other one *is* open, but maybe it should have at least the same
> priority as this report had?

They have the same severity. priority is up to the assignee.
Comment 14 Martin Seidler 2010-12-20 18:56:10 UTC
(In reply to comment #12)
> > maybe someone could edit the list for the "Most annoying bugs" since M4, too?
> > 
> > See:http://en.opensuse.org/openSUSE:Most_annoying_bugs_11.4_dev#openSUSE_11.4_Milestone_4_of_6
> 
> It's a Wiki used and written by our community. Could you take care of that as
> part of our community ? Thanks a lot!

Surely. ->
http://en.opensuse.org/index.php?title=openSUSE%3AMost_annoying_bugs_11.4_dev&action=historysubmit&diff=31640&oldid=31544


(In reply to comment #13)
> > Sorry, the other one *is* open, but maybe it should have at least the same
> > priority as this report had?
> 
> They have the same severity. priority is up to the assignee.

Thanks. Sorry for asking so much. So my hopes will relay on:
"Assignee:  	E-mail List <kernel-maintainers@forge.provo.novell.com>"