Undervolting the X41 2525-6NU

Message

billyb · #1 Post by **billyb** » Thu Sep 22, 2005 4:48 pm

In spite of searching backwards quite a ways, I didn't find any postings with hard numbers for undervolting the X41. So I thought I would post the results of my experiments here for others to try.

After a week plus of testing voltage settings and stability on many, many nights, here are the settings I am currently using with Centrino Hardware Control Ver 1.9 Beta 2:

06X - 0.700V
08X - 0.716V
09X - 0.732V
10X - 0.748V
11X - 0.764V
12X - 0.796V
14X - 0.860V
15X - 0.908V

This has resulted in CPU temp drops of between 5 degrees C at 6X to around 12 degrees C at 15X from the normal voltages, and not insignificant indications of increase in battery life (between an extra 30 and 45 mins as displayed in the battery icon on the taskbar, since I did not actually do full battery rundown tests).

Methodology used - at each multiplier, starting with 15X, use CHC's short test to reduce voltage to the point where the machine locks up and refuses to run. Then back off one step and run Prime95 Ver 24 in Torture Test mode using the Blend option. Find a voltage at which Prime95 runs for at least 8+ hours with no errors. Then back off one more step for safety and set that as the voltage for that multiplier. I guess if you want to be ultra cautious, you can back off two steps rather than one, but I have applied these settings to 3 different X41s, all 2525-6NUs. (The later addition of a 1 GB Transcend SO-DIMM into each of these machines seems to have made no difference, all 3 machines passed the Prime95 test again for 2 hours at each multiplier step.)

For example - 15X. With voltage set to 0.844V, CHC short test locks up. Reboot. Set CHC to use next higher voltage step, 0.860V, CHC short test passes. Prime95 shows errors in less than 2 mins. Set CHC to use next higher voltage step, 0.876V, CHC short test passes. Prime95 runs for 8 hours, no errors. Back off a further one step, 0.892V. Set that as voltage for 15X.

For multipliers of 8 and 9, actually, Prime95 ran perfectly for 8 hours plus at the lowest voltage of 0.700V, but I set it slightly higher simply because the steps were available. This also gave me the confidence to leave the 6X setting at the very lowest of 0.700V.

For whatever it's worth, I have also enabled CHC's Custom Dynamic Switching, though it doesn't really appear to make much difference.

It's well over a week now since the three machines have been set to use these voltages, and so far there has been no issue of any kind with these settings.

If anyone has any comments or suggestions on these settings or the methodology used to derive them, please let me know here. I would presume the X41 Tablet should work with the same settings, but as don't have a Tablet you would have to try it yourself and verify. Actually it's probably better to apply the method to each machine rather than simply plug in the numbers as listed here, as there will probably be variations in tolerances across CPUs from different production batches.

billyb

Weymouth_Stan · #2 Post by **Weymouth_Stan** » Fri Sep 23, 2005 8:58 am

On first reading, your post was as indecipherable to me as a tax-return in Sanskrit, so I googled 'CHC' and now I'm having fun

. I love to tweak. My CPU temp is down to 50c using your voltage figures.
If anyone else wants to try underclocking/volting, here's a link to CHC:
http://www.pbus-167.com/chc.htm

Yotam · #3 Post by **Yotam** » Fri Sep 23, 2005 11:02 am

Good for you guys. I tried to use these values and my X41T crashed twice.

Anyway, I set it for only 2 multipliers -
X6 - 0.7V (The lowest, and it works)
X15 - 0.972V

(When on AC power I want it to have enough power for better performance, while on battery I want it with much battery life and less heat)

Lets see when will it crash...

ThinkNick · #4 Post by **ThinkNick** » Wed Sep 28, 2005 1:35 pm

To chime in here, I got similar / same results as Yotam on my X41 tablet 1866-6GU.

X6 was very stable at 0.7V (will future vers of CHC let this go lower?), while X15 worked fine at 0.956V, though using your methodology of backing off one step I would come out with 0.972V as Yotam reported.

Quite a change!

Yotam · #5 Post by **Yotam** » Wed Sep 28, 2005 2:30 pm

ThinkNick wrote:To chime in here, I got similar / same results as Yotam on my X41 tablet 1866-6GU.

X6 was very stable at 0.7V (will future vers of CHC let this go lower?), while X15 worked fine at 0.956V, though using your methodology of backing off one step I would come out with 0.972V as Yotam reported.

Quite a change!

Hi ThinkNick,

Just to let you know, I tried to undervolt again after my first crash shock, and now I'm working on 0.956V with no problems. Try it. If it crashes (may crash after the test, while working on the computer) use the 0.972V, although there is not much difference.

amgdoc · #6 Post by **amgdoc** » Mon Nov 28, 2005 10:44 pm

just tried it out. really helps keep the fan quiet.

amgdoc · #7 Post by **amgdoc** » Tue Dec 06, 2005 5:54 pm

just wanted to add that on the 1.2ghz processor on the x40, the stable voltage for the lowest is 0.796 (6x). anything lower seems to make the computer freeze, especially after hibernating.

(also, the temp is a nice 32 degrees C, no fan noise at all!!!! )

EDIT: change voltage to higher level from 0.748 (6x) since notebook became unstable.

bha* · #8 Post by bha* » Wed Dec 07, 2005 3:35 pm

billyb

Thanks for the tip. Values listed work perfectly on my X41 252563G.

jonnnny · #9 Post by **jonnnny** » Mon Apr 17, 2006 10:46 am

Thank you so much!!

I had tried looking for something in regards to undervolting my x41 but couldn't find anything so I assumed it wasn't possible (since it is a low voltage processor?) And lord knows I dont have the patience to do all those tests.

But I tried these values and everything is perfect!
normal of 44 degrees down to 38

THANKS AGAIN!

edit: some more stats
(by the way this is a 6LU)
battery life... maxed out at 6:30 on an 8-cell
not sure how accurate that is, haven't actually timed it, but WOW crazy...

KentM · #10 Post by **KentM** » Mon Apr 24, 2006 10:06 am

amgdoc,
What temperature do you get with the updated voltage on your LV 1.2 GHz?

I see limited change in temp when I change voltage on my 1.2 CPU. You reported 32C which is way lower than I ever had (normally 39C). I seems I can't affect the temp/CPU fan at all even if I apply very aggresive settings. Is there a commit button I haven't discover?

Has anyone found the lowest settings for a LV 1.2 GHz X40? and also for a 2.0 GHz X32?

Kent

mhca · #11 Post by **mhca** » Sun Jun 25, 2006 7:54 am

I am just ending my last Prime tests on my X41.

These are the voltages I have tested stable.

06X - 0.700V

08X - 0.716V

10X - 0.780V

12X - 0.844V

15X - 0.940V

I haven't tried to push them down even further because I want to feel confident that my system is stable.

I'm looking forward to enjoying a more silent and less power consuming laptop

mhca · #12 Post by **mhca** » Tue Jun 27, 2006 6:09 am

I am wondering if it is possible to have a program which only manages the voltages and displays CPU speed in system tray while I can still use IBM's Power Management software?

I would like to use my different schemes which also controls LCD brightness and I also like the fact that it doesn't fully charge my batteri if it is only discharged to 96%.

Edit: At first IBM's power manager just didn't seem to want to power up but now it's no problem.

Undervolting the X41 2525-6NU

Undervolting the X41 2525-6NU

Who is online