Another Intel Linux Power Regression Is Being Investigated

[phoronix]

Phoronix: Another Intel Linux Power Regression Is Being Investigated

Power regressions are still easy to come by with the Linux kernel and other areas of the open-source stack... Multiple users have been reporting of a recent power increase on newer versions of the Linux kernel, which seem to track down to the Intel i915 DRM driver...

http://www.phoronix.com/vr.php?view=MTc3NTE

[mcloud]

Beat 'em up and find the true regression first!

[liam]

I'm curious to see what results Michael finds. Is this an issue just with apple or is it general to haswell?

[Michael]

I'm curious to see what results Michael finds. Is this an issue just with apple or is it general to haswell?

So far I have some 3.15 <-> 3.16 power regression confirmed for non-Haswell, non-Apple hardware... That appears to be on the CPU scheduler side, but still investigating.

[liam]

So far I have some 3.15 <-> 3.16 power regression confirmed for non-Haswell, non-Apple hardware... That appears to be on the CPU scheduler side, but still investigating.

Ugh. Neverending story:/
Are these all Intel?

[matt_g]

If you read further into the mailing list thread:

> Actually, with an image of white-noise displayed fullscreen while
> powertop takes a 20 second measurement, the idle power consumption > shoots up to 11W, with FBC enabled. This experiment actually
> invalidates my guess #3. The white noise image will not compress much
> at all, while my typical test screen is a couple of static
> black-and-white terminal windows (filling up the screen), which will
> compress well.
>> So it would appear 4W is the actual cost of having a full-size 20MB
> framebuffer on this system.

I guess the monster resolution just really hurts w/o fbc. My hsw has
1920x1080 panel which, assuming the same refresh rate, means your
display refresh requires 2.5x the bandwidth mine does.

At least from my reading of it the increase in power consumption is mostly due to the stupidly high resolution of Mac laptops - morale of the story is if you want to use such a large resolution expect more power consuption regardless of whether you turn on Frame Buffer Compression or not.

Or to put it another way "1920x1080 ought to be enough for everybody..."

[ihatemichael]

The i915 driver is completely BROKEN on Arch Linux (Linux 3.16.1-1) and it has been broken for days now.

https://bbs.archlinux.org/viewtopic.php?id=185650
https://bugs.archlinux.org/task/41045
https://bugs.archlinux.org/task/41573

This broke OpenGL, Xv (mpv/mplayer/vlc) and other stuff such as KWin compositing.

There has been some patches sitting around on freedesktop.org for days now and it's sad that no single developer even bothered to review them. Leaving all the users with broken drivers and no GPU support for days now.

Why is it taking so long? Having the GPU completely unusable for days is pathetic and unacceptable.

[    7.130035] [drm:i915_gem_init] *ERROR* Failed to initialize GPU, declaring it wedged

Damn it.