Watt-Creep: ~120W->250W, CPU/GPU < %5

[rover7]

Over the last month, my watt-usage at ~idle (with apps running, but not very active) has gone from:

~120W->250W, CPU/GPU < ~5% (according to process explorer).

I can't seem to bring it back down to anything close to where it was, using the same set of apps, more or less.

I'm using a watt-meter at the plug to see watt usage.

Any suggestion to figure out why the seemingly large increase in power consumption without much if any increase in CPU/GPU usage?

I think it's possible that even legitimate apps I'm using now are using more power than before, but shouldn't such a remarkable change in wattage be reflected in CPU/GPU usage? I.e, what system resources are being used to use all those extra watts?

System specs:
=
- Windows 7 Home Premium 64
- i7 920 @2.67MHz
- NVIDIA GeForce GTX 285
- RAM was 6B (3x2GB, speed unknown). Added 12GB (3x4GB 1600MHz). Working in tri-channel. 'My computer' reports 18GB total, 16GB usable.

 

[logainofhades]

The lower wattage is possibly when speedstep kicks in and lowers the clock speed and voltage of your cpu. If speedstep kicks off and the cpu ramps back up to full speed, that would cause an increase in wattage. Turn off speedstep in the bios then boot up and check if it fluctuates like that.

 

[rover7]

logainofhades,
I updated/edited my post to reflect that my wattage has gone up, but I can't seem to bring it back down, even if I set max CPU to 1% in Windows Power Settings. Do I need to enable speedstep in BIOS to enable that CPU-max Windows Power Setting? I haven't checked yet in BIOS whether this is enabled or not. Any way to do so without rebooting?
Thanks!

 

[logainofhades]

Yes, you have to reboot to get into the bios.

 

[rover7]

BIOS showed Intel SpeedStep "enabled".

Any other ideas as to why my comp is using so many watts at idle CPU/CPU?

I set min/max CPU to 1/2% in "Power Saver" power mode on this desktop. That seemed to have no effect.

Anything else I might try?

 

[popatim]

disable aero desktop

 

[rover7]

disable aero desktop

I already did that popatim.

Any other ideas? How could my machine use 250+ watts in 95% CPU idle?

 

[rover7]

Actually, I separated out my router. Previously, it was using just ~30 watts max but someone else started using it.

My desktop now "only" uses 160 watts at 97% idle.

 

[logainofhades]

Given your system's stats, 160w is not bad. i7 920's and GTX 285 are quite power hungry.

 

[rover7]

Ok.

I'm curious to understand why starting e.g. Chrome for the first time adds 30 Watts permanently, even after I close Chome.

I see 2 chrome.exe procs using 3.5% GPU each.

When I close chrome, that GPU usage goes away and no other procs except for csrss.exe (uses 0.7%), but the watts go down by only ~3 if anything.

I'd suspect chrome initialization starts up some e.g. system tasks that eat up CPU or GPU even after it is finished, however, I don't see any GPU usage (according to list of procs in ProcessExplorer) and CPU doesn't go up to make up for anything the GPU was previously doing.

I'll have to double check what CPU is before ever starting Chrome after a reboot to see how much CPU overall, goes up, but doesn't come down, after starting and ending Chrome.

If you know how I might avoid a reboot and clean up any mess starting Chrome might have left, I'd be happy to learn it! : )

Thanks!

 

[nukemaster]

250 seems very high for idle. I think my 920 system idled about 150.

Get CPU-Z and GPU-Z to see if one or both cpu and video card are getting locked at a higher clock speed.

http://www.cpuid.com/softwares/cpu-z.html

http://www.techpowerup.com/downloads/SysInfo/GPU-Z/

I know Nvidia had an issue that caused the 650 ti to get stuck at a high speed in Media Center(but that only added 20-30 watts). Maybe you have something similar.

Image of the driver issue
http://imageshack.us/a/img211/3241/nvidiamediacenter.png

 

[rover7]

Again, turned out it was actually not using as many watts as I thought. The router watt-use went up, and I hadn't expected that. I suppose other users were hitting it hard. I should have separated the router in my test but neglected to. Sorry for all the commotion. However, all the info presented here was helpful. Thanks everybody!