I'm not sure if this was intentional or not, but Liquorix 3.2.0-5 uses 25W more power (from the wall) than 3.2.0-4 under load on my system. Also, I'm not sure if this is related, but my system decided to reboot itself today for the first time. The CPU may be generating a little more heat, but nothing to cause any problems.

Here are my specs:
Linux Mint KDE 12, Intel i5 2500k @ 4.7GHz, Nvidia GTX 480 SLI, 2 x 2GB DDR3 OCZ Reaper HPC 1866MHz CL7, OCZ Vertex 3 SSD

Let me know if I can do anything to help. The power issue isn't too much of a concern especially if speed has been improved as a result, but I fear that some instability has been introduced as well.[/list]
Back to top

I'll be pushing a new kernel today that fixes this. In the meantime, you can boot to an older kernel or use the nothreadirqs parameter from grub.
Back to top

Cool. I just wanted to make you aware or it. I have been using -4 for the time being. Cheers.
Back to top

Looks like this was fixed in the subsequent kernel, but broken again in the latest one.
Back to top

Please test 3.2.0-6.dmz.1.
Back to top

Here is what I tested:
3.2.0-4.dmz.1 = 255W
3.2.0-5.dmz.1 = 285W
3.2.0-5.dmz.2 = 255W
3.2.0-6.dmz.1 = 285W

Does that help?
Back to top

Are these wattages being measured off the wall with a real watt meter, like kill-a-watt? I'd never trust any internal watt measuring algo in the operating system.
Back to top

Yes, they are measured from the wall with a Zalman power meter/fan controller. Readings are shown on an LCD screen, not in software. I've never had this problem with any of the other liquorix kernels. My measurements are taken at 100% CPU load, 0% GPU load. Also, the kernels that use more power seem to introduce slight instability though it could be my imagination.
Back to top

It really sounds like your GPU driver isn't being installed correctly. If your cpu load is 100% and your gpu load is 0%, then faulty power management on the gpu can increase your power draw from the wall. Especially with your setup, SLI requires two or more cards to render something in concurrency. I can imagine that getting sloppy with power management, even if you're not using the cards to do anything, can cause serious power issues.

Look, this is the diff between the configuration of 5.dmz.2 and 6.dmz.1 on 64-bit (32-bit being identical except the 'AACRAID' module is set to M).



Can you show us the output of inxi -bxx?
Back to top

What were the changes between -4 and -5? It's possible that I chose the wrong kernel in GRUB by mistake in one test. I know power usage has been right around 255W under 100% CPU load on this box since I built it. It only jumped up to 285W when I installed 3.2.0-5+. When I switch back and forth from -4 to -5, power draw goes from 255W to 285W consistently. The nvidia driver is the same.

I've actually installed several different versions of Ubuntu, Linux Mint, and Debian with kernels 2.6.32 - 3.2.0 in the past couple months. Different nvidia driver versions were used. In all cases, power consumption was right around 245W-260W. This is the only case that I've seen it spike significantly.

GPU usage is actually *near* 0% since I don't do any gaming when booted into Linux. However, it does fluctuate slightly with 2D rendering as you'd expect. My cards are just way overkill for simply scrolling and moving windows around. Compositing is disabled because it doesn't work with xinerama and multiple screens.

Here is the output you requested:


Don't trust the CPU info. It is a 4-core clocked at 4.7GHz.
Back to top