Notebook Power Test Results

More popular than ever, but some are still very noisy.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Avalanche
Posts: 137
Joined: Mon May 29, 2006 6:29 am
Location: IN, USA
Contact:

Notebook Power Test Results

Post by Avalanche » Mon Jul 09, 2007 5:44 am

My Dell Inspiron 8600 is wrapping up it's 4th(!) year of service as I type this, and just a few weeks ago I found some excellent free software tools to improve its battery life, lower operating temperatures and quiet the cooling fans down. Since I am a reasonably well read computer enthusiast, and I only found these programs recently, I thought it might be fair to guess not too many people are aware of them.

This thread introduced me to Notebook Hardware Control, and I forget where I learned about Dell Inspiron Inspiron/Latitude/Precision fan control. I didn't write these programs, nor am I the first to use them, but I did some minimally scientific tests using them, and made some charts in Excel, so I thought I'd share the results.

Specifications:
Dell Inspiron 8600
Pentium M 1.4GHz, Banias
1.25 GB DDR
855PM

The first thing I wanted to try was lowering the CPU voltage. Since this option was unavailable in the BIOS, and since Dell (AFAIK) hasn't released any tools for doing this, I thought it wasn't possible. With NHC, it is VERY possible. :-) Just run NHC, and check under the "CPU Voltage" tab.

The Pentium M can (as most of you know) Speed Step. This chip runs at 1.4 GHz when loaded, and 600 MHz at idle. The Dell factory voltages were:

1400 MHz - 1.484 V
600 MHz - 0.940 V

Using CPU Burn-in I ran the core at maximum load for 20 minutes and marked the final temperature reported by the Fan Control program. Then I lowered the voltage, and repeated. Results are shown below:
Image

As you can see, the notebook ran fine all the way down 1.052 V. Any lower and Windows froze up intermittently. This is 70% of the stock voltage! I was suprised myself. The temperatures also dropped significantly.

The blue line indicates relative power usage by the CPU, calculated as,
Image
where this article and others suggest CPU power usage is proportional to voltage squared. So the final result shows that CPU power usage is nearly halved by using the new core voltage!

Using my handy Kill-A-Watt,
Image
I measured my total power usage. The laptop was plugged in, the screen was set to full brightness and CPU Burn-in was used again.
Image

For reference turning the display off resulted in -10 W used, and turning the display from maximum brightness to minimum brightness reduced energy usage by 5 W.

So in conclusion, I now use NHC to set the CPU voltages as specified above, and Dell fan control to leave the fan off when CPU temp is below 55ºC. When heavy processing is happening, the core will heat up to 55ºC then the fan kicks on low, and the core cools down in a minute or two back to 45ºC. These setting result in much quieter and cooler computing than before. I haven't tried to run any battery lift tests, but it seems like the battery run-time must be significantly improved by the new voltage settings.

I hope this report is helpful and encourages someone else to try these two programs!

Beyonder
Posts: 757
Joined: Wed Sep 11, 2002 11:56 pm
Location: EARTH.

Post by Beyonder » Mon Jul 09, 2007 8:32 am

I'm a big fan of NHC, although admittedly I've never done any testing with the kill-a-watt to see how much I can reduce my CPU usage by.

One question: how did you determine your laptop was stable at those voltages? I think you may have to do more than just boot into windows to determine stability.

Beyonder
Posts: 757
Joined: Wed Sep 11, 2002 11:56 pm
Location: EARTH.

Post by Beyonder » Mon Jul 09, 2007 9:11 am

I'll conduct similar tests on my laptop when I get a chance. Out of curiosity, do you have the spreadsheets for doing the calculations? I'll just add my data to your sheet.

thx.

Avalanche
Posts: 137
Joined: Mon May 29, 2006 6:29 am
Location: IN, USA
Contact:

Post by Avalanche » Mon Jul 09, 2007 11:21 am

Beyonder wrote:how did you determine your laptop was stable at those voltages?
Good point. I know a lot of overclocking websites use a battery of tests, but I mostly just did my usual work on it. At first I was a notch lower (1.036 V) but I kept having freezing problems. One bump up and they went away.
Beyonder wrote:Out of curiosity, do you have the spreadsheets for doing the calculations?
The spreadsheets really aren't much, but if you're interested I could clean 'em up and upload them. The only calculation is that V^2/1.484^2. I'll try to do that tomorrow.

cansan
Posts: 127
Joined: Thu Oct 28, 2004 1:33 pm
Location: Germany

Post by cansan » Mon Jul 09, 2007 4:39 pm

I have a Dell D610 that I have been running undervolted for about 2 years now with NHC.

Mine is running at 1.148v at 1.86G, and 0.7v at 0.8G. At 1.86G, it shaves off 6 watts at full load, and about 1-2? watts at full load at 0.8G. You can monitor power drain if you are running on battery power in NHC.

To find out how far you can undervolt, start with the default voltage, drop a little, run Prime 95 or some other stress test and see if you have errors. If not, lower some more and so on... until you have some errors, then put it back up 1-2 notches. You may want to load the gpu while testing, because for example on mine, I can go lower without the gpu loaded, I guess because of less heat.

Overall, a great way to save some battery power and a cooler laptop.

Avalanche
Posts: 137
Joined: Mon May 29, 2006 6:29 am
Location: IN, USA
Contact:

Excel Spreadsheet

Post by Avalanche » Tue Jul 10, 2007 9:47 am

Hey Beyonder,

My Excel Spreadsheet is available here. If you took the time to run a similar test, it'd be neat to see how well the two results line up. I was pretty surprised to find the nearly linear temperature/voltage correlation.

Let me know if there's any problems.

BTW, I did the temperature test with the fan set to low. I don't really recommend doing it with no cooling. Forgot to mention that above. :shock:

Beyonder
Posts: 757
Joined: Wed Sep 11, 2002 11:56 pm
Location: EARTH.

Post by Beyonder » Tue Jul 10, 2007 10:06 am

Cool, I'll mess around with it tonight. I have a junky Toshiba laptop that my work gave me that I currently run NHC on, so it'll be an interesting test.

FlorisNielssen
Posts: 141
Joined: Fri Aug 10, 2007 1:33 pm
Location: The Netherlands
Contact:

Post by FlorisNielssen » Sun Aug 12, 2007 4:47 am

Nice program, that NHC! Been playing with it today and came across Ati's PowerPlay feature, which underclock's your videocard. Unabling it (finally!) solved a problem I always had with with Maple. It ran terrible, mouse trembled over it, maple-screen locked. All over now!

Also lowerd my CPU voltage from 1.2625V to [s]1.08751V[/s] 1.1V and saw my CPU temperature (under load by 2 Prime95 test's) lowered from 65C (where te fan began to spin at maximum speed) to [s]50C[/s] 52C.
Too bad my T2500 doesn't allow the CPU voltage to be lower than 0.95V, the default with a 6x multiplier. Does anybody know if this can be overruled?

By the way, avalanche, have you tried running Prime95 with 1.052V? I bet it'll give lots a errors... If Windows just hangs if you lower it móre I can't believe it if it doesn't. 1..875V gave an error with me, so I had to higher the CPU voltage. But Windows didn't hang for a second by then.
It doesn't have to be a problem the CPU makes some mistakes now and then:
Prime95 STRESS.txt wrote:CAN I IGNORE THE PROBLEM?
-------------------------

Ignoring the problem is a matter of personal preference. There are
two schools of thought on this subject.

Most programs you run will not stress your computer enough to cause a
wrong result or system crash. If you ignore the problem, then video games
may stress your machine resulting in a system crash. Also, stay away from
distributed computing projects where an incorrect calculation might cause
you to return wrong results. Bad data will not help these projects!
In conclusion, if you are comfortable with a small risk of an occasional
system crash then feel free to live a little dangerously! Keep in mind
that the faster prime95 finds a hardware error the more likely it is that
other programs will experience problems.

The second school of thought is, "Why run a stress test if you are going
to ignore the results?" These people want a guaranteed 100% rock solid
machine. Passing these stability tests gives them the ability to run
CPU intensive programs with confidence.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Sat Aug 25, 2007 8:25 pm

RMClock is another excellent mobile CPU tinker software. I'm amidst testing a Pentium M Banias 1.7 @ ULV levels. Great little CPU, if a little dated, and a great program.

Post Reply