Bug 67321

Summary:	[dpm] rs880: Power usage doesn't drop with dpm=1
Product:	xorg	Reporter:	Dominik Kopp <suse>
Component:	Driver/Radeon	Assignee:	xf86-video-ati maintainers <xorg-driver-ati>
Status:	RESOLVED MOVED	QA Contact:	Xorg Project Team <xorg-team>
Severity:	enhancement
Priority:	medium
Version:	unspecified
Hardware:	x86 (IA32)
OS:	All
Whiteboard:
i915 platform:		i915 features:

Description Dominik Kopp 2013-07-25 19:33:19 UTC

My laptop (Lenovo Edge 11) has the Mobility Radeon HD 4200 Series chip (RS880M) 
(windows says it's a HD 4225)

It consume with fglrx ~16W. (kernel 3.4.47)
With radeon (kernel 3.11-rc2) and power_profile=low ~20W
With radeon.dpm=1 (kernel 3.11-rc2) ~20W

linux-7goe:/home/dom # echo battery >/sys/class/drm/card0/device/power_dpm_state
linux-7goe:/home/dom # echo low> /sys/class/drm/card0/device/power_dpm_force_performance_level 
linux-7goe:/home/dom # modprobe ec_sys
linux-7goe:/home/dom # cat /sys/kernel/debug/dri/64/radeon_pm_info 
uvd    vclk: 0 dclk: 0
power level 0    sclk: 20000 vddc_index: 1

The fglrx still uses ~4W less than radeon.dpm=1

It would be great when radeon.dpm=1 uses as less power as the fglrx driver.

Comment 1 Alex Deucher 2013-07-25 20:11:22 UTC

Please(In reply to comment #0)
> My laptop (Lenovo Edge 11) has the Mobility Radeon HD 4200 Series chip
> (RS880M) 
> (windows says it's a HD 4225)
> 
> It consume with fglrx ~16W. (kernel 3.4.47)
> With radeon (kernel 3.11-rc2) and power_profile=low ~20W
> With radeon.dpm=1 (kernel 3.11-rc2) ~20W

The power usage between DPM and profile=low should be pretty comparable on these chips.  DPM just has the advantage of being able to dynamically change the clocks and voltage transparently when performance is required vs. having to change it manually with the profile code.

Comment 2 sotiris papadimitriou 2013-07-30 22:17:52 UTC

I have the same graphics card and the difference compared with fglrx is huge, indeed. 
With DPM=1 or low profile or fglrx the clocks are adjusted on 200MHz but temperature is + ~6°C on radeon and energy consumption on + ~4W. 

Will DPM ever reach the levels of fglrx ?

Comment 3 Alex Deucher 2013-07-30 22:32:16 UTC

(In reply to comment #2)
> Will DPM ever reach the levels of fglrx ?

It should be pretty close on most asics.  I'm not sure what else fglrx does on rs780 off hand.  Unfortunately, the chip is pretty old so it's harder to find docs or people the remember all the details.

Comment 4 D Rhodes 2014-02-26 18:36:27 UTC

(In reply to comment #3)
> (In reply to comment #2)
> > Will DPM ever reach the levels of fglrx ?
> 
> I'm not sure what else fglrx does on rs780 off hand.


A suggestion: Stutter Mode

I'm not a linux expert but I've done some ad hoc testing. The results imply that stutter mode is working on fglrx but not working on r600g.

My hardware is Athlon II x2 240, RS880G + 128MB sideport RAM
Kernel 3.12.9

At idle with the HT bus set to its default 2GHz the wallplug power is:

fglrx 31.7W

r600g 36.2W

If I set the HT bus to 400MHz, in the BIOS, as a crude approximation of disconnecting the HT bus 80%, I get:

fglrx 31.2W

r600g 32.7W (a reduction of 3.5W)


I read that, when using UMA, it should be possible to disconnect the HT bus 93% of the time, and presumably even more with sideport RAM.

If my idea is plausible, is it possible to enable stutter mode in r600g?

Comment 5 Martin Peres 2019-11-19 07:41:49 UTC

-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/xorg/driver/xf86-video-ati/issues/73.

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.