Bug 23262 - ATI Radeon Mobility KMS modeset=1 is slower than modeset=0
Summary: ATI Radeon Mobility KMS modeset=1 is slower than modeset=0
Status: RESOLVED INVALID
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/Radeon (show other bugs)
Version: 7.4 (2008.09)
Hardware: All Linux (All)
: high major
Assignee: xf86-video-ati maintainers
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-08-11 22:08 UTC by Bryce Harrington
Modified: 2016-02-25 21:23 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments
results.html (9.47 KB, text/html)
2009-08-11 22:08 UTC, Bryce Harrington
no flags Details
Xorg.0.log.mode1 (151.06 KB, text/plain)
2009-08-11 22:11 UTC, Bryce Harrington
no flags Details
dmesg (45.31 KB, text/plain)
2009-08-11 22:11 UTC, Bryce Harrington
no flags Details
lspci-vvnn (10.71 KB, text/plain)
2009-08-11 22:12 UTC, Bryce Harrington
no flags Details
Xorg.0.log from modeset=1 (98.44 KB, patch)
2009-08-12 18:42 UTC, Jerry Amos
no flags Details | Splinter Review
dmesg from the T40 (46.36 KB, text/plain)
2009-08-12 18:45 UTC, Jerry Amos
no flags Details

Description Bryce Harrington 2009-08-11 22:08:41 UTC
Created attachment 28531 [details]
results.html

Forwarding this bug report from Ubuntu reporter Jerry Lamos.

[Problem]
Performance regression when using the KMS -ati driver corresponding to version 1:6.12.99+git20090805.bd03977e along with kernel, mesa, and libdrm from the kms ppa: https://edge.launchpad.net/~ubuntu-x-swat/+archive/kms.

[Original Report]
Performance is actually worse than -vesa on this system.

Measurements are from IBM Thinkpad T40 with ATI Radeon Mobility 7500.  See attached file for specific data.

Col 1 is measurements with xorg driver vesa.
Col 2 is nomodeset no KMS xorg ati driver 6.12.99+git.....
Col 3 is modeset=1 KMS xorg ati driver 6.12.99+git.....
Comment 1 Bryce Harrington 2009-08-11 22:11:09 UTC
Created attachment 28532 [details]
Xorg.0.log.mode1

I understand this is a known issue, so haven't forwarded other performance bugs, however this one has some crunchy data to go with it, so am forwarding for tracking purposes.
Comment 2 Bryce Harrington 2009-08-11 22:11:36 UTC
Created attachment 28533 [details]
dmesg
Comment 3 Bryce Harrington 2009-08-11 22:12:50 UTC
Created attachment 28534 [details]
lspci-vvnn
Comment 4 Michel Dänzer 2009-08-12 03:17:17 UTC
The log file shows that the X driver lacks KMS support, so it fails to enable the DRI (and uses the hardware behind KMS' back...). If there is a problem when the X driver actually uses KMS (and higher AGP transfer rates don't help), please reopen with the corresponding log file.
Comment 5 Jerry Amos 2009-08-12 18:42:18 UTC
Created attachment 28574 [details] [review]
Xorg.0.log from modeset=1
Comment 6 Jerry Amos 2009-08-12 18:44:39 UTC
Results from GtkPerf on IBM Thinkpad T40 1.5 gHz P4 ati radeon mobility 7500
All times are in seconds with higher number being slower.

GtkPerf0.40Aug1117:35:092009				
karmic.A3.radeon.Mobility7500.1:6.12.99+git20090805.bd03977e-0ubuntu2				
[CODE]				
		    jaunty	A3_mode0	Modeset=1
GtkEntrytime:____________		0.13	0.07	0.10
GtkComboBoxtime:_______		1.78	2.33	8.87
GtkComboBoxEntrytime:___		1.08	1.56	4.86
GtkSpinButtontime:_______		0.32	0.37	2.96
GtkProgressBartime:______		0.86	0.36	5.12
GtkToggleButtontime:_____		0.35	0.46	3.05
GtkCheckButtontime:______		0.31	0.26	0.97
GtkRadioButtontime:______		0.60	0.40	1.31
GtkTextViewAddtexttime:__		1.29	1.64	1.75
GtkTextViewScrolltime:____		0.51	0.49	1.60
GtkDrawingAreaLinestime:__		1.74	1.01	2.67
GtkDrawingAreaCirclestime:		1.34	1.31	2.55
GtkDrawingAreaTexttime:__		1.78	8.00	1.89
GtkDrawingAreaPixbufstime:		0.44	0.77	3.01
				
Total	time:	12.54	19.04	40.72
[/CODE]				
Comment 7 Jerry Amos 2009-08-12 18:45:54 UTC
Created attachment 28575 [details]
dmesg from the T40
Comment 8 Jerry Amos 2009-08-12 18:52:33 UTC
Ctrl-Alt-F1 switches quickly to the typical modeset=1 small font 48 line screen on this 1024x768 lcd, and lots of characters per line?  Then pressing Ctr-Alt-F7 switches quickly right back to this graphics screen.  There doesn't appear to be any reset.  It's almost like flipping windows.  Yes there are some complaints in syslog but isn't that always the case with linux.

Jerry
Comment 9 Bryce Harrington 2009-08-13 11:04:22 UTC
[Jerry has attached the log files from his more recent tests with kms, showing the regression.]
Comment 10 Christopher M. Penalver 2016-02-25 21:23:07 UTC
Jerry Lamos, Ubuntu Karmic reached EOL on April 30, 2011. For more on this, please see https://wiki.ubuntu.com/Releases .

If this is reproducible on a supported release, it will help immensely if you filed a new report with Ubuntu by ensuring you have the package xdiagnose installed, and that you click the Yes button for attaching additional debugging information running the following from a terminal:
ubuntu-bug xorg

Also, please feel free to subscribe me to it.

For more on why this is helpful, please see https://wiki.ubuntu.com/ReportingBugs.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.