The Notebook Undervolting Guide

[996GT2]

I found this guide posted by flipfire at Notebookreview forums to be extremely helpful in undervolting the Penryn T8300 in my Thinkpad T61.

Original VCore was 1.1375v at 2.4 GHz. After undervolting and stress testing with Orthos, I found that my system was stable at 0.9250v all the way to the 10x multi and stable at 0.9375v for the 11x and 12x multipliers.

My idle temps are now in the low 30C range, and my load temps in Orthos are in the 50-51C range. Not bad at all for a notebook CPU!

EDIT: Screenshots and detailed info lower in thread:

This should encourage you to try it out.

Screenshot taken with WiFi on, 35% brightness, Performance on Demand mode, and Vista's "Balanced" power setting. I had been using the laptop for about 30 minutes prior to taking the screenshot, so battery if not full.

My T61 has a 2.4 GHz T8300, 3 GB RAM, NVS 140M Graphics, and a 160GB 7200RPM HDD. Not the most power-saving combination of components, but the battery life definitely saw significant improvements after undervolting.

 

[thecoolnessrune]

Thanks for the link. Followed it and was running 1.275V for 2Ghz. Now doing .9750V for 2Ghz, .9250V for 1.8Ghz, and .9000V for 1.6Ghz. .9V is as low as it will go in the list so I didn't bother with running it at lower multipliers since that would just be less efficient wouldn't it?

Processor would peak around 85 Celsius under load now does about 75-76. Idles at around 36-37 down from 43. Its probably lower in laptops with integrated graphics. I have an 8400GS that idles at 56 Celsius. I have an HP laptop and its a dv6000t model but apparently the service tag isn't on recall. So bad or not I'm kinda stuck with it

 

[996GT2]

Well processor power usage is not simply determined by voltage. It's also affected by clock speed. So if you have a 1 GHz CPU at .9v, it's still using less power than a 2 GHz CPU at .9v. If you enabled the SuperLFM and other low multiplier modes, you could probably squeeze a bit more battery life out of your laptop.

 

[thecoolnessrune]

Originally posted by: 996GT2
Well processor power usage is not simply determined by voltage. It's also affected by clock speed. So if you have a 1 GHz CPU at .9v, it's still using less power than a 2 GHz CPU at .9v. If you enabled the SuperLFM and other low multiplier modes, you could probably squeeze a bit more battery life out of your laptop.

I see then, spose I'll enable all the other multipliers at .9000 volts then.

 

[xSauronx]

Originally posted by: 996GT2
I found this guide posted by flipfire at Notebookreview forums to be extremely helpful in undervolting the Penryn T8300 in my Thinkpad T61.

Original VCore was 1.1375v at 2.4 GHz. After undervolting and stress testing with Orthos, I found that my system was stable at 0.9250v all the way to the 10x multi and stable at 0.9375v for the 11x and 12x multipliers.

My idle temps are now in the low 30C range, and my load temps in Orthos are in the 50-51C range. Not bad at all for a notebook CPU!

how did it affect your battery times?

 

[996GT2]

Originally posted by: xSauronx

Originally posted by: 996GT2
I found this guide posted by flipfire at Notebookreview forums to be extremely helpful in undervolting the Penryn T8300 in my Thinkpad T61.

Original VCore was 1.1375v at 2.4 GHz. After undervolting and stress testing with Orthos, I found that my system was stable at 0.9250v all the way to the 10x multi and stable at 0.9375v for the 11x and 12x multipliers.

My idle temps are now in the low 30C range, and my load temps in Orthos are in the 50-51C range. Not bad at all for a notebook CPU!

how did it affect your battery times?

I'm getting about 20 more minutes, or about 4.5 hours on my T61 (9 cell battery) with wifi on, screen at 35% brightness, and the "balanced" power setting. I'd say that's pretty good for a laptop with discrete graphics and a 7200 RPM hard drive.

The other benefit is that my CPU runs about 10-15C cooler at both idle and load, so I rarely hear the fans kick up in normal use.

 

[thecoolnessrune]

With my HP dv6500 (15" Screen, 8400GS wifi left on and a 5400RPM drive) on Power Saver mode and a 12 cell battery, I'm getting around 5 and a half hours of battery life. That's pretty damn stellar and lets me get through my entire college day on a single battery charge.

And, like 996, I'm running about 10-15C cooler, which makes it a lot more bearable in my lap and for the noise in the classroom.

 

[996GT2]

Originally posted by: thecoolnessrune
With my HP dv6500 (15" Screen, 8400GS wifi left on and a 5400RPM drive) on Power Saver mode and a 12 cell battery, I'm getting around 5 and a half hours of battery life. That's pretty damn stellar and lets me get through my entire college day on a single battery charge.

And, like 996, I'm running about 10-15C cooler, which makes it a lot more bearable in my lap and for the noise in the classroom.

I just set a new battery time record for my T61...when I was watching the Olympics track/field finals last night in a dark room, I had the screen dimmed all the way and wifi on (to check stats on some of the athletes). It was idle about 1/3 of the time and I was just looking at Wikipedia for the rest, but it gave an estimated 5 hours 14 minutes of battery life on my 9 cell...after an hour, the battery gauge was still dead on at just over 4 hours.

This is a HUGE improvement to many on the Thinkpad forums who report 3.5-4 hours of battery life with a T61 in similar configuration, but not undervolted.

And the other benefit is that undervolting actually prolongs the life of the processor and many other internal components due to the reduced heat output.

 

[WannaFly]

I just got a M1330 with the T8300 and started to do the undervolting thing, but got confused when it said " If you have SuperLFM or IDA i suggest leaving this out for the meantime. See down below later on for what these features do."

Maybe I'll try it again.

 

[996GT2]

Originally posted by: WannaFly
I just got a M1330 with the T8300 and started to do the undervolting thing, but got confused when it said " If you have SuperLFM or IDA i suggest leaving this out for the meantime. See down below later on for what these features do."

Maybe I'll try it again.

You don't need to leave SuperLFM out...I have the box checked in RMClock and it works fine. The SuperLFM mode is a "low frequency mode" that halves your FSB speed at the lowest 6x multiplier, giving your CPU a 600MHz operating speed at idle conditions to prolong battery life.

The SuperIDA mode is a high performance mode that temporarily raises the clock on one core when that core is under heavy load. I would leave this unchecked because it's hard to stability test this mode (you can't "Lock" into this mode like you can with a specific multi).

What I did was checked all the boxes from SuperLFM to 12x multi, with 0.9250v (the lowest allowable by the CPU) for all multipliers up to 10x and 0.9375v for 11x and 12x multipliers. Orthos failed after 6 hours on .9250v for those higher multipliers, so I went the safe route and increased voltage a bit for the last two.

I'd be happy to help if you have questions since I have a T8300 in my Thinkpad also.

 

[WannaFly]

996GT2,
Tried it again just now. Without any tweaks and with orthos running full, my high cpu temp is 55, if i change anything in that program and run it, within seconds my cpu is up to 65-70. I unchecked SuperLFM and IDA, left all the default voltages, etc. When I uncheck "auto-adjust", none of the temperatures change.

Can you tell me on the "CPU Info" tab does it recognize your core, mine says "Unknown core".
Could you provide some screenshots? :-D

 

[996GT2]

Originally posted by: WannaFly
996GT2,
Tried it again just now. Without any tweaks and with orthos running full, my high cpu temp is 55, if i change anything in that program and run it, within seconds my cpu is up to 65-70. I unchecked SuperLFM and IDA, left all the default voltages, etc. When I uncheck "auto-adjust", none of the temperatures change.

Can you tell me on the "CPU Info" tab does it recognize your core, mine says "Unknown core".
Could you provide some screenshots? :-D

Here you go:

http://i78.photobucket.com/alb...20/996gt2/RMClock1.jpg
http://i78.photobucket.com/alb...20/996gt2/RmClock2.jpg

DO NOT use Auto-Adjust. Start by ONLY checking the 12x (highest) multiplier, then selecting a low voltage such as .9375v to start. Then go to the "Maximum performance" profile and check only the 12x multiplier box (scroll down). Right click the Orthos icon in your taskbar and select the Max Performance Profile. Then go into OS settings (next to the CPU settings screen, button is on top) and pick a setting (I use "balanced".

Now, go into Windows Power options (the battery icon in taskbar)->more power options->change plan settings (for whatever plan you're using)->advanced settings->processor management, and set BOTH the minimum and maximum states to 100% for battery and plugged in. This will allow RMClock to take control of your CPU Voltages.

Run Orthos with your CPU LOCKED into the 12x multi and see if it's stable for one hour at 2.4 GHz and the low voltage you selected. If it is, then you can automatically apply that voltage to all of the lower voltages as well and then you can use the "Performance on Demand" profile (make sure you check all the boxes from 6x to 12x). This profile gives you speedstep, but you first have to test what voltage your CPU needs to run stably at 2.4 Ghz.

 

[The Mailman]

thanks for all the info, this will be very handy come the school year!

 

[996GT2]

Np, good luck!

Edit: Screenshot of battery life after 1/2 hr use (conditions described in OP)

http://i78.photobucket.com/alb...996gt2/BatteryLife.jpg

 

[coolVariable]

Originally posted by: 996GT2

Now, go into Windows Power options (the battery icon in taskbar)->more power options->change plan settings (for whatever plan you're using)->advanced settings->processor management, and set BOTH the minimum and maximum states to 100% for battery and plugged in. This will allow RMClock to take control of your CPU Voltages.

can you explain what this is for?
I did not see this in the NBR forums guide.

 

[WannaFly]

996GT2, thanks for the screenshots.. It looks like without any change in settings, with orthos running, my laptop was maxing out at 1.2Ghz for some reason, Afer applying the performance on demand its not running at 2.4, which explains the higher temps. It's up to about 77 after ~10 minutes of running orthos, where as withour RMClock it was only at about 55, but running at 1.2Ghz (i think).

Another weird thing, my voltages only go to 9.5, no lower for all multipliers *shrug*

 

[coolVariable]

Originally posted by: WannaFly
996GT2, thanks for the screenshots.. It looks like without any change in settings, with orthos running, my laptop was maxing out at 1.2Ghz for some reason, Afer applying the performance on demand its not running at 2.4, which explains the higher temps. It's up to about 77 after ~10 minutes of running orthos, where as withour RMClock it was only at about 55, but running at 1.2Ghz (i think).

Another weird thing, my voltages only go to 9.5, no lower for all multipliers *shrug*

That is the same thing for me. Temps with RMclock are much higher due to the CPU cconsistently clocking higher.
I enabled throtteling and now the temps seem more in line with what I had before.
I also cannot go below 9.5.


What CPU do you have?
I have a Merom L7400 (1.5GHz Core 2 Duo)

My temps are:
RMClock at 9.5v with throttling enabled: 56C-60C (normal windows)
RMClock at 9.5v without throttling enabled: 90+C for 100% CPU load
no RMClock: 80C for 100% CPU load (but seems to step down the multiplicator automatically)

 

[996GT2]

Originally posted by: coolVariable

Originally posted by: WannaFly
996GT2, thanks for the screenshots.. It looks like without any change in settings, with orthos running, my laptop was maxing out at 1.2Ghz for some reason, Afer applying the performance on demand its not running at 2.4, which explains the higher temps. It's up to about 77 after ~10 minutes of running orthos, where as withour RMClock it was only at about 55, but running at 1.2Ghz (i think).

Another weird thing, my voltages only go to 9.5, no lower for all multipliers *shrug*

That is the same thing for me. Temps with RMclock are much higher due to the CPU cconsistently clocking higher.
I enabled throtteling and now the temps seem more in line with what I had before.
I also cannot go below 9.5.


What CPU do you have?
I have a Merom L7400 (1.5GHz Core 2 Duo)

My temps are:
RMClock at 9.5v with throttling enabled: 56C-60C (normal windows)
RMClock at 9.5v without throttling enabled: 90+C for 100% CPU load
no RMClock: 80C for 100% CPU load (but seems to step down the multiplicator automatically)

You mean .95v right?

Also, 9.5x multi is CORRECT for your CPU, since you have an L7400. 9.5x166=approx 1.5 GHz.

I have a T8300, so 12x200 FSB=2.4 GHz

To make your voltages to all the way, you need to go into advanced CPU settings in RMClock and select "Mobile" instead of "Desktop." Mine defaulted to "Desktop" also.

Your temps should not be that high at all for .95v. I think your heatsink may be mounted wrong or you may have a ton of dust in your laptop.

Also, you need to set CPU to 100% in vista for the power settings so RMClock can control your CPU Voltage. Try it without doing this, and you will see in your CPU monitor that your CPU is still using default voltages, not the ones you set in RMClock. The clock speeds are controlled either way, so the CPU does not stay at 100% since it's governed by RMClock.

996GT2, thanks for the screenshots.. It looks like without any change in settings, with orthos running, my laptop was maxing out at 1.2Ghz for some reason, Afer applying the performance on demand its not running at 2.4, which explains the higher temps. It's up to about 77 after ~10 minutes of running orthos, where as withour RMClock it was only at about 55, but running at 1.2Ghz (i think). Another weird thing, my voltages only go to 9.5, no lower for all multipliers *shrug*

You need to enable "Mobile" CPU in RMClock. See a few lines up in my post. The lowest possible for a T8300 is .9250v. The CPU is hard coded to not use anything lower.

Perhaps your CPU was locked at 1.2 because you were in power saver mode. In RMClock, you need to select the multipliers that go with each power profile as well as select an OS profile. For Max performance, only check the 12x multi. For power saver, only check the 6x multi. For Performance on Demand, check all but the IDA. Then, set your OS power scheme for the Max performance profile to "High performance," "balanced" for "performance on demand," and so on.

MAKE SURE YOU GO INTO ADVANCED CPU SETTINGS AND DO WHAT I STATED IN BOLD A FEW POSTS UP!

Your temps seem high...maybe try blowing out the dust as well?

 

[coolVariable]

Sorry - I meant 0.95v with that.
Multiplicators are 6x, 7x, 8x, 9x

I did not change the Vista power settings to 100% yet.
What does this do?
Despite not changing this, RMClock reports that the CPU is using the lower voltages.

I have NO dust and the laptop seems as warm as always (had it now for a year).
Could these temps be normal for a tablet/this CPU?
Otherwise, RMClock / HWMonitor might be reporting wrong temps?


I noticed another issues with this: when running RMClock, I notice stuttering when playing online videos. When I exit RMClock the videos run smoothly.

 

[996GT2]

Originally posted by: coolVariable
Sorry - I meant 0.95v with that.
Multiplicators are 6x, 7x, 8x, 9x

I did not change the Vista power settings to 100% yet.
What does this do?
Despite not changing this, RMClock reports that the CPU is using the lower voltages.

I have NO dust and the laptop seems as warm as always (had it now for a year).
Could these temps be normal for a tablet/this CPU?
Otherwise, RMClock / HWMonitor might be reporting wrong temps?


I noticed another issues with this: when running RMClock, I notice stuttering when playing online videos. When I exit RMClock the videos run smoothly.

When you leave the vista power settings stock, the Vista power management will conflict with RMClock because both will be trying to set CPU voltages, multipliers, etc.

This is why you need to set them to 100%, so that only RMClock can control the voltages.

The stuttering might be due to the fact that your CPU is not stable at a super low voltage. Since you already have an LV chip, it might not go much lower.

Try the steps in the guide for testing stability. Remember, your CPU has to pass at least one hour of Orthos Low FFT at a certain voltage before you can move on to a lower voltage.

 

[coolVariable]

Originally posted by: 996GT2

Originally posted by: coolVariable
Sorry - I meant 0.95v with that.
Multiplicators are 6x, 7x, 8x, 9x

I did not change the Vista power settings to 100% yet.
What does this do?
Despite not changing this, RMClock reports that the CPU is using the lower voltages.

I have NO dust and the laptop seems as warm as always (had it now for a year).
Could these temps be normal for a tablet/this CPU?
Otherwise, RMClock / HWMonitor might be reporting wrong temps?


I noticed another issues with this: when running RMClock, I notice stuttering when playing online videos. When I exit RMClock the videos run smoothly.

When you leave the vista power settings stock, the Vista power management will conflict with RMClock because both will be trying to set CPU voltages, multipliers, etc.

This is why you need to set them to 100%, so that only RMClock can control the voltages.

The stuttering might be due to the fact that your CPU is not stable at a super low voltage. Since you already have an LV chip, it might not go much lower.

Try the steps in the guide for testing stability. Remember, your CPU has to pass at least one hour of Orthos Low FFT at a certain voltage before you can move on to a lower voltage.

I changed the power settings for Vista.
For plugged in (what I am running currently) they already were set to 100% both.

I still have the problem with the stuttering video. It only happens maybe every 20-30 seconds and it just pauses for maybe half a second. But it is definitely worse than without RMClock.
Watching RMClock this doesn't tax the CPU much and it stays on a 6x multiplier (which is set already at the stock Voltage of 0.95v ... that remains unchanged so it must be RMClock that is causing the stuttering, not the CPU or the undervolting!).
Having multiple windows open, outlook running, etc. the current temperature reads 61C.

 

[996GT2]

Does it stutter when you are locked into 6x multi without RMClock? It probably wouldn't stutter if it was running full speed when you aren't using RMClock

 

[coolVariable]

Originally posted by: 996GT2
Does it stutter when you are locked into 6x multi without RMClock? It probably wouldn't stutter if it was running full speed when you aren't using RMClock

Never stuttered before.
I am not limiting the multiplicator with RMClock - it is set to performance on demand.

 

[WannaFly]

Well, afte messing with it most of the day i've seems to lower the voltages, I'll be running overnight test the next few nights - but i ran each one for 45min-1hr and it worked fine.

My (hopefully) only problem now, because i cant adjust it lower then .95, which iswhat superlfm was set to use, it doest really affect my idle temps. They are about the same whether I am running RMClock, or without. It DOES affect whn in use, i see almost a 10C difference at full load. Hopefully it will positively affect battery life also.

996GT2, thank for helping me earlier. What program are you using to measure your temperatures?

 

[996GT2]

I just use the default monitoring in RMClock. It seems pretty accurate. You can download CoreTemp to double check.