[Bug 110865] Rx480 consumes 20w more power in idle than under Windows
bugzilla-daemon at freedesktop.org
bugzilla-daemon at freedesktop.org
Mon Aug 26 21:45:45 UTC 2019
https://bugs.freedesktop.org/show_bug.cgi?id=110865
--- Comment #22 from Dieter Nützel <Dieter at nuetzel-hh.de> ---
Hello Alex and Martin,
I've tried both on my
Polaris 20, RX580 8 GB Sapphire Technology Limited Nitro+ Radeon RX 580
- v2 patched into amd-staging-drm-next (before inclusion of v3)
- v3 with amd-staging-drm-next
https://cgit.freedesktop.org/~agd5f/linux/commit/?h=amd-staging-drm-next&id=ca82748783d8189a54a85f2ea1c2710182ba6138
Both flicker with green/black (?) horizontally lines over both screens.
Mostly during power level switch. For example during mouse movement/interaction
(wheel) and mouse pointer traverse from konsole/etc. to desktop (KDE5 Plasma
5.xx, here).
UVD load (mplayer etc.) is not enough to fix it.
E.g. radv (vkcube) not.
But other gfx load (vkmark/glmark2, etc.).
When there is lower gfx demand during the above tests (glmark2 -b buffer) the
flicker came up, again.
Martin's observation
[-]
If i do
"echo low > /sys/class/drm/card0/device/power_dpm_force_performance_level"
the flickering stops.
So the flickering is caused by the automatic powermanagement / reclocking.
[-]
Works here, too (tested with v3).
But I never could go below ~32 W !!!
Tested with both Nitro+ BIOS modes.
The PSTATE_xxxx wouldn't change on my card. They stay @ 600/1000 all the time!?
GFX Clocks and Power:
300 MHz (MCLK)
300 MHz (SCLK)
600 MHz (PSTATE_SCLK)
1000 MHz (PSTATE_MCLK)
750 mV (VDDGFX)
32.76 W (average GPU)
GPU Temperature: 31 C
GPU Load: 0 %
MEM Load: 3 %
Any hints?
And sorry for my bad English this time - my best friend from beginning of
German Gymnasium died after 6 years of fight against cancer. He aged only 52.
Leaving a wife and two little girls...
--
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/dri-devel/attachments/20190826/b9717bbc/attachment-0001.html>
More information about the dri-devel
mailing list