How far are you undervolting your Pentium-M processor?

NFS4

No Lifer
Oct 9, 1999
72,636
47
91
I just started dabbling with undervolting my Pentium-M 740 (1.73GHz) last night.

Intel lists voltages at 1.356 @ 1.73GHz and 0.988 @ 800MHz for my processor.
Centrino Hardware Contol lists 1.308 and 0.988 at the respective speeds.

I undervloted to 1.260 and 0.956 with no ill-effects so far.

That being said, how far have you guys gone down and is there any real tangible battery life increase by dropping the voltage a little?
 

The Linuxator

Banned
Jun 13, 2005
3,121
1
0
I will be able to verify that once my Thinkpad arrives in a couple of days, but IMHO there should be a very tangeible battery life increase I mean, we are talking about a
53.75 % clock cycle decrease, and a 32 % VCPU decrease theoretically given that everything other than the cpu is running at the same specs, you should be experiencing a 20-30 % increase in battery lifetime, haven't you tested how much battery lifetime you gained in this process ? Also if you have a GPU of modifiable VGPU such as new ATI one, you should be able using this uttility ( which I beleive is the one you are using), to make your laptop even more energy conservant, but I have to test it out on my laptop once I get the chance. ;)
 

Trippytiger

Senior member
Mar 3, 2005
410
0
0
I'm down to 0.700V at 600MHz, down from the stock 0.988V. At 1700MHz, I'm at 1.068V. It could be slightly lower, but I had some stability problems.

My friend says he can do 1600MHz at 0.9V. I hate him. :p
 

GnomeCop

Diamond Member
Jun 17, 2002
3,863
0
76
havent tried any of this undervolting stuff yet, gonna have to investigate when I get home.
 

Accord99

Platinum Member
Jul 2, 2001
2,259
172
106
I think pretty much all of the Dothans are good for 0.7v at the lowest multiplier, though because of Dothan's excellent leakage characteristics you won't see much of a increase in battery life since it's dominated by the rest of the laptop components. The power savings can be pretty significant at high CPU loads though. My stable results for 730 P-M is 0.7v at 800MHz, 0.956v at 1.6GHz.
 

Conroy9

Senior member
Jan 28, 2000
611
0
0
I assume this is completely safe to do, since I can't think of how undervolting could damage anything?

Are there any reasons not to try this?
 

makken

Golden Member
Aug 28, 2004
1,476
0
76
I have a P-M 1.86Ghz, current lowest and highest settings are:
0.700V @ 6x (800Mhz)
1.032V @ 14x (1.86Ghz)

When I get home, I can post my full range. I bet i can go down one more step at 10x, 12x ant 14x, I know I get stability problems at 10x at 0.796V, but 14x at 0.988V passed a 30 minute stability test wtih no problem
 

Mingon

Diamond Member
Apr 2, 2000
3,012
0
0
My 1.6dothan 400 fsb is running at 2.133 using a wire mod in an aopen sonoma laptop. At 2.13 I am totally stable at 1.21v :) the lowest I can go down to at 6 x 133 (800mhz) is 0.754. I use CHC which is very good and has a stability tester (albeit short) built into it.
 

Conroy9

Senior member
Jan 28, 2000
611
0
0
I guess I need to answer my own question to get a serious answer..

Before doing any important work at a lower voltage, make sure you test your machine thoroughly. I found a stable setting last night of 1.036v@2ghz, and ran Prime95 overnight. It survived with no problems, so I bumped it up to 1.052v and ran Prime95 all day while I was doing my normal work. After around 8 hours, Prime95 got an error, so now I'm trying again at 1.068V.

It looks like it would be hard to figure out what point is completely "stable"; if it can run for 15 hours before erroring, it could also conceivably run for a week before the error, etc..
It also might be hard to notice that the cpu is producing errors unless you're constantly running stress tests.

I'm still researching to see whether there's any possible way this could damage the system.

The temperature drop is pretty amazing though.
 

vegetation

Diamond Member
Feb 21, 2001
4,270
2
0
Been stable for many hours under prime95 with 1.036 with a 16x multiplier with my 1.6 dothan. I figure the best way would be to up it a notch anyways just to ensure stability.
6x multiplier at .7v no problem. Overall I can't really tell much, but I think it's running a little cooler on ac since just scrolling a complex webpage can trip the multiplier to full throttle for brief amounts of time and I guess all that adds up during web browsing sessions.
 

Conroy9

Senior member
Jan 28, 2000
611
0
0
Nice article!
When you have tijme, you might want to update it with the instructions for using the new interface of RMClock 1.6 and 1.7. This link shows the changes.

Also, there might be one more thing to watch out for when undervolting. If you're messing with the fan speed as well, make sure to monitor the HD temp also to make sure the HD isn't overheating while the fans are off because of a cool CPU.
 

makken

Golden Member
Aug 28, 2004
1,476
0
76
Alright, I just looked up my full range in CHC:
I'm running a 533Mhz FSB Dothan, 14x max (1.86Ghz)

6x (800Mhz) @ 0.700v
8x (1.06Ghz) @ 0.716v
10x (1.33Ghz) @ 0.844v
12x (1.60Ghz) @ 0.942v
14x (1.86Ghz) @ 1.020v

I tested the system with both CHC's built in 15minute full stability check, as well as running Prime95 overnight.

14x @ 1.004v threw a BSOD after about an hour
14x @ 1.020v ran Prime95 fine for over 6 hours.
 
Feb 24, 2001
14,513
4
81
Mine is overclocked from 1.6 to 2.13, I'm assuming that's why it requires more voltage.

AS5, and a Targus cooler with 2 fans

0% load, 800mhz, 0.700v, 25C
100% load, 800mhz, 0.700v, 27C

100% load, 2.13ghz, 1.212v, 39C

I could undervolt down another notch or two, but get errors in Prime 95 after 1 hour. At 1.212v, will run with no problems.

So not quite as low as others, but I attribute it to a 33% overclock.
 

Mingon

Diamond Member
Apr 2, 2000
3,012
0
0
Originally posted by: BrunoPuntzJones
Mine is overclocked from 1.6 to 2.13, I'm assuming that's why it requires more voltage.

AS5, and a Targus cooler with 2 fans

0% load, 800mhz, 0.700v, 25C
100% load, 800mhz, 0.700v, 27C

100% load, 2.13ghz, 1.212v, 39C

I could undervolt down another notch or two, but get errors in Prime 95 after 1 hour. At 1.212v, will run with no problems.

So not quite as low as others, but I attribute it to a 33% overclock.


Mines the same as yours although I cant go lower than .75 @ 800mhz. Mind you I run considerably hotter as my notebooks heatsink/pipe is a piddly little thing - 64c sometimes. Mind you my 1.6@2.133 in my ct-479 runs considerably warmer than that even with a zalman 7700 i.e. idle 40c tops at 50c.
 

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
102,395
8,558
126
i looked into undervolting my notebook after reading the advanced battery guide at laptoplogic, but i have a different version of the i855 chipset (the one without integrated graphics) so i wasn't sure if the clocking/volting program would work. anyone know if it does? dell inspiron 600m here
 

Accord99

Platinum Member
Jul 2, 2001
2,259
172
106
Originally posted by: Viper0329
How much battery life does this actually save though?

Not much at low CPU usage, since the P-M is barely using any power in those situations and the other laptop component's power usage becomes the dominant factor in battery life. Maybe another 10%. For high cpu usage, ther should be a significant gain, upwards of 50% if you don't have a power hungry video card. But it's rare that a laptop would be under continuous high CPU loads while running off the battery.
 

makken

Golden Member
Aug 28, 2004
1,476
0
76
not that much, I think i only see a 10 or so minute increase.

I do it mainly for the heat though. without underclocking, at full load, my cpu temp tops out at nearly 80C, with underclocking, it gets reduced to high 60's, low 70's
 

meksta

Senior member
Jul 24, 2001
252
0
0
I am running 0.7v at 600mhz on my T42. It extended my battery by about 30 minutes....

which is significant cuz I can barely run 3 hours now.