Bug 93471 - [radeonsi] power consumption on pitcairn card stays high after 2. monitor disconnected
Summary: [radeonsi] power consumption on pitcairn card stays high after 2. monitor dis...
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/Radeon (show other bugs)
Version: git
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: xf86-video-ati maintainers
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2015-12-21 20:19 UTC by Malte
Modified: 2016-01-05 07:03 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
drm part of dmesg (6.78 KB, text/plain)
2015-12-21 20:19 UTC, Malte
no flags Details
Xorg.log startup with second monitor attached (66.45 KB, text/plain)
2015-12-22 09:02 UTC, Malte
no flags Details
Xorg.log startup and xrandr commands executed (47.48 KB, text/plain)
2016-01-02 13:19 UTC, Malte
no flags Details

Description Malte 2015-12-21 20:19:55 UTC
Created attachment 120641 [details]
drm part of dmesg

When starting the PC with a single monitor, the power consumption is about 60W (was 50-55W with a HD4650 before).

cat /sys/kernel/debug/dri/64/radeon_pm_info
uvd    vclk: 0 dclk: 0
power level 0    sclk: 30000 mclk: 15000 vddc: 900 vddci: 850 pcie gen: 2

When adding a second monitor the power consumption stays at this level until its get activated.
Once activated (xrandr --output HDMI-0 --auto), the consumption ramps up to 79W.

cat /sys/kernel/debug/dri/64/radeon_pm_info
uvd    vclk: 0 dclk: 0
power level 0    sclk: 30000 mclk: 120000 vddc: 900 vddci: 1000 pcie gen: 2

Disabling the second monitor again (xrandr --output HDMI-0 --off), does not reduce the consumption again. Not even unplugging helps.
While I read that its simply not possible to get the low power with two active monitors, could it be possible to manually or
automatically switch to the previous clock if all but one monitor have been disabled?
Moreover could the signal simply be cloned without consuming +29W if resolution and refresh rate are equal?
The second monitor is a projector permanently connected, but switched off 90% of the time.

Xorg: From Debian 8.1 stable
Kernel: 4.3.3
Card (bought as R7 370): VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Curacao PRO [Radeon R9 270] (rev 81)
Comment 1 Michel Dänzer 2015-12-22 01:22:28 UTC
Please attach the corresponding Xorg log file.
Comment 2 Malte 2015-12-22 09:02:34 UTC
Created attachment 120649 [details]
Xorg.log startup with second monitor attached
Comment 3 Michel Dänzer 2015-12-22 09:13:36 UTC
Any chance you could try the current xf86-video-ati release 7.6.1? http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/commit/?id=ddaba449e8d6fe9fc0d97085e4045843fd8d7af9 might help.
Comment 4 Malte 2015-12-24 11:43:07 UTC
The first try of compiling with the script failed due 32/64bit multiarch. I will continue after xmas.
Comment 5 Malte 2016-01-02 13:19:32 UTC
Created attachment 120766 [details]
Xorg.log startup and xrandr commands executed

I managed to compile and run xorg somehow (got xdm to run, but no default login or preferred window manager).
Executing xrandr --output HDMI-1 --off a first time properly reduces power, however the main screen goes blank for a short moment and new console windows appear in xdm (as if another session would have been restored)
Executing xrandr --output HDMI-1 --auto increases power again. No changes in xdm
Executing xrandr --output HDMI-1 --off a second time, does nothing. No xdm changes, no power changes. Exiting and restarting xorg, I can repeat this procedure with the same result.
Comment 6 Malte 2016-01-02 16:26:58 UTC
As the mentioned fix is in the ATI driver, it would be good if I actually could build and use it. Expected it to be in the radeon/radeonsi driver and concluded the missing xf86-video-ati wont be a problem.

==  Processing:  "driver/xf86-video-ati"
==        configuration options:
make: *** Keine Targets angegeben und keine »make«-Steuerdatei gefunden.  Schluss.
build.sh: "make " failed on driver/xf86-video-ati



***** Failed components *****
    driver/xf86-video-v4l
    driver/xf86-video-ati
Everything else could be build.
Comment 7 Malte 2016-01-03 08:56:55 UTC
I used the source from Debian, added the source of the mentioned patch and build a new .deb packet. Now
xrandr --output HDMI-0 --off
reduces power as expected, thanks for pointing to the patch :)
But the question remains if it is possible to get the lower power consumption even with two screens (showing the same picture).
Comment 8 Michel Dänzer 2016-01-05 07:03:59 UTC
(In reply to Malte from comment #7)
> I used the source from Debian, added the source of the mentioned patch and
> build a new .deb packet. Now
> xrandr --output HDMI-0 --off
> reduces power as expected, thanks for pointing to the patch :)

No worries, thanks for testing. Resolving this report as fixed.


> But the question remains if it is possible to get the lower power
> consumption even with two screens (showing the same picture).

If both screens are using HDMI connections and the exact same timings, using the same CRTC for both may work.


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.