The "Undervolting" Guide

0.125 + 0.125 = 0.25

If a bluescreen happens it will just restart, and none of the unstable settings will be saved.

 

Oh I Feel So Stupid For Not Noticing This


Thank You !!!

 

I tried the undervolting (I have a T8300)
after undervolting to 1.000V, and any more undervolting gave me a BSOD. okay, so I restart, leave it at 1.000V

but now when i check my CPU, the core clock fluctuates from 1.2GHz to 2.4GHz and back and forth. wat the hell, or is that just the norm that i never took notice of?

 

okay i actually found out why it has such a low clock
it's cause the current multiplier is 6.0x... now how do i make it so that the default is 12.0x?

 

You need to check the 13x multiplier, and undervolt it. If you leave the 13x unchecked, your CPU will throttle back to 12x instead of going right to 12.5x.


In Advanced CPU Settings, you can have "Engage Intel Dynamic Acceleration (IDA) unchecked.

 

No, that's intended behavior though.

THat's "Performance on Demand" in action: It ramps up your CPU when it needs the processing power, and drops it down into low frequency mode when idling to conserve power and reduce heat.

Liken it to an engine: Normal, "Performance" mode is like keeping your engine running at high RPM even when stopped. This gives you instant power when you want it (by popping the clutch!) but it causes your engine to run really hot and waste gas. "Performance on Demand" allows your engine to idle when not needed. It'll run up to full speed in a faction of a second when the CPU needs to do something. You still get power when you want it, and saves you "gas" when you don't.

 

Is the performance not crippled by disabling the IDA ? I mean Intel must have made for a reason ?

 

The reason we leave it unchecked is:
1) that it only boosts the multiplier by 1, so for the T8300, it goes from x12 to x13
2) it seems to be buggy (can't remember where I heard this)
3) it hardly ever comes on so there's no point in worrying about it

 

Okay!
Another question.. if I get to lets say 1.0000v on my highest multiplier without BSOD while stresstesting, then I should be able to set the lower multiplier to at least as low as the highest multiplier (in this case 1.0000v), right ?

 

that would be correct.

 

Sorry for the inactivity here, i have been researching on how to undervolt on the new montevina platform.

-Click Defaults
-Tick Auto-adjust intermediate VID's
-Choose 1v for your 13x mutliplier (Press the up and down key and watch all multipliers voltages move automatically)
-Click Apply

All done

 

I dont know why I didn't undervolt sooner. So far I have took 10C off @ 1.1000v from 1.2500v. I think I can get a lot more out of it, seeing as my CPU is only 1.46GHz.

Before:


After:

 

I'm trying to undervolt my T8300 on an m1330 after reading this guide, but when I apply the stress test and go to "CPU info" in RMClock, the Current state is ALWAYS 6.0 x multiplier, with 1.0 Voltage, even though the maximum multiplier is 12.0x. Even if I put other stress on the system, it still show a 6.0x multiplier. I tried unchecking all the boxes that were not 12x in the Main profile page, and nothing. I even went as far as undervolting the 12x multiplier to .9250 V (the lowest possible), and I tried the stress test, and didn't even get a BSOD. This makes me think that the CPU never throttles up for some reason. Anyone got any advice? This only seems to happen if I open RMClock.

 

I hit 0.95v with the 11x multiplier, it has taken a massive 21C off my load temp.



I cant even feel any warm air coming out of the vents now. Thanks for the tutorial.

 

Go to the 'Performance on demand' Sub-profile page and make sure the indexes there are all ticked.

 

I did, and still nothing. Whenever I open RMClock, the multiplier becomes locked to 6x (even when I completely exit out of the program afterwards). CPU-Z shows that it stays at a 6x multiplier too, and stays that way until I reboot. After that, the CPU throttles up and down like it should, until I open RMClock again... weird . I was looking forward to trying undervolting too...

 

-Close RMclock

-Run RMwipeout.reg in the RMclock folder and carefully follow the guide again from scratch.

If you have any other power managment utilities turn them off. I dont see why it shouldnt work for your m1330

 

can someone tell me that if I uninstall the RM Clock tool, the voltages will go back to default?

If so, does it mean that everytime your laptop starts up, the software "overrides" the voltages according to your settings. Which means, you always need the software running in background to keep those low voltages. Its not like you use the software to permanently set low voltages. Am I right?

 

based on my understanding, you are correct. you will need rmclock running in order to have the voltage changes to take effect.

 

Hooray, that reg file seems to fix it. Thanks! Running stress tests for my undervolt now. I'd rep you for that, but I already rep'd you for the guide when I read it .

 

IDA is not high performance.

What IDA does is shuts down one of your CPU's two cores (thus 1/2 normal power) then overclocks the other core by 1x. This is only useful for an app running on a single thread (think non-windows based software) with NO background processes running at all (windows background processes run in seperate threads on your other core, even if you are running an app that only utilizes one core on it's own.

Thus, IDA is virtually never used, and on the odd chance it is (typically only in odd, situations) it has marginal impact.

It's really just junk.

 

I just want to say thank you, you got me from 1.7500V to 1.000V safely on a T7250 CPU.

Thanks!!

 

Thanks for the guide. I'm down about 10C over stock. Repped.

 

Hello,

My company bought me a HP notebook 6710b powered by C2D T7250 @ 2 Ghz. I want to undervolt it and I have just discovered this great thread!

Tolkann, you said you have successfully undervolt your T7250. Could you please post the voltages you uses for each P-State in order for me to start my undervolting with your settings as basis?!

Thank you very much!

Have a nice day!
Steve

 

Be aware that Tolkannns voltages might not be stable for yours. Your CPU's might have different voltage tolerances. Its best you find the optimal voltage yourself.

 

I got a question Flip, HWmonitor and RMclock are reporting different temps, with HWmonitor's being 4-5C higher...err...which one is right?

 

Thanks for the advice, flipfire

 

I have a Dell XPS 1330 with a T7250
I undervolted with RMclock and here are the results :

6x ~ 8x @ 0.90 V
9x @ 0.9125 V
10x @ 0.95 V

Stress tested all the multipliers individually in orthos for around 20 mins. I am surprised at the huge jump in voltage in the 10x multiplier compared to 9x

My temps :

CPU :

Idle : Earlier 48C Now 38C
Max : Earlier 74C Now 51C

GPU :

Idle : Earlier 67C Now 55C
Max : Earlier 106C Now 81C


I have also done the copper mod on the gpu. I removed the stock thermal paste on the CPU and applied arctic cooling MX-2 on both the CPU and the GPU.

 

Thanks akg7091 for your experience with T7250

 

Use Coretemp as a third check.

RMclock might have it wrong if you have a newer revision of a chip

but then again, HWmonitor has also been wrong in the past for certain chips

 

Just to be clear, since my T8300 has no half-multiplier (max is 12x), ticking off IDA should be of no consequence at all, right?

 

Exactly, I've got the exact same CPU, however, I now decided that I'd leave IDA on and bump up the voltage on it by a step.

 

this guide is nice, thank you haha but ocing is fun

 

Correct. IDA is purely for single-thread operation, and will virtually never be used in Windows. It's only an issue with the hidden (in RMClock) half multipliers for the T9300.

 

hey i found some interesting things, this will definately help some achieve a better/lower/cooler system

check out the post.

essentially, rmclock/coretemp post hypothetical voltages and CPU-z reports actual volts.

http://forum.notebookreview.com/showthread.php?t=229282&page=174
first post by me, urbanos

 

Your incorrect about cpu-z

CPU-z has always reported the correct voltages, in sync with RMclock. But after version 1.41, it started to display incorrect temps for Merom/Penryn processors. I believe its showing desktop equivalent voltages of the Merom/Penryn CPU's.

Coretemp and RMclock are correct. There is only one VID, the one RMclock requests for.

Attached is a screenshot of my Pentium M 725 w/ CPU-z 1.45 . Notice how the voltage are in sync? This is also the same for my Yonah cpu. Its only Merom and Penryn that was affected after v1.41

 

Indeed. I recently went through this a while back in this thread: CPU-Z reports incorrect temps and voltages for my T9300 penryn processor too. Confirmed with other apps (everest 'n such) that RMClock is correct, and CPU-Z is not.

 

Hmm, what should I enable and disable on the "advanced cpu settings" page? What are all those C1E, C2E etc. ? And what about all the stuff under "automatic thermal protection"?

I have a T9300 CPU

 

THIS STILL HASN"T BEEN STICKIED??!!?!?

lol

 

Dont touch the sleep states, they should automatically be set.

Just disable IDA on that page

Not worth a sticky, undervolting isnt for everyone. I would have stickied it myself if it really was

Its sub-stickied into the hardware booth

 

does this work fine with montevina cpus as well? Great program guys.

 

No im afraid. RMclock doesnt support the new Montevina platform.

The RMclock development team has died out.

 

does undervolting effect the way the laptop charges?

also does the method described here work with the intel GM965 chipset?

 

To the first, not dramatically - it will charge slightly faster while your operating it as the processor is using somewhat less power(making more available to charge the battery), but probably not a noticeable difference.

 

howdy folks, I am trying to undervolt a lenovo t61p with a t9300. I'd dropped it to 1.1375, 1.1, 1.05, 1.0, 0.9875 and finally to 0.975 where it seemed to be running stably (orthos ran for about 15m)

When I took it to 0.9625 orthos failed and there was a bluescreen.

After rebooting my computer, orthis is now reporting a failure at the default 1.1375V! It seems to consistently error within 60 seconds of starting (often within 5-6).

I tried deleting the configuration files (prime.ini, local.ini) for orthos and re-running but get the same results.

Did I break something?
Copy of some runs of the splog.txt available here

 

No you just didnt do it correctly.

Go to the RMclock folder and run RMwipeout.reg. This will flush all the settings

Restart your notebook

 

My T8300 is now undervolted at 1V at the 12th multiplier, max effective temp, 55C. Wuosh

 

thanks flipfire that did it.

and thanks x 2 for the guide in general, very helpful. I've gotten it down to 0.975 and life is good (or something)

 

Hi!
I try to undervolt my CPU but the program writes 1.200V to all my multipliers...is that right? Oh and I have Core 2 Extreme X9000.

 

Hey guys!

I just got my new laptop (LG p300) with a T8300.
I read and followed the guide and it worked beautifully (12x @ 1.0125v).

However, I decided to read some other forums too and see if there are any other approaches and I found this one (sorry if some one mentioned this before me, I didnt want to read through 120 pages!):

With a T8300 for example:
index 0 with 12x / SuperLFM (in my case at 0.925v)
index 7 with 12x / normal (in my case @ 1.0125v)

So that means im either running at 1.2 Ghz with 0.925(!) v or 2.4 Ghz with 1.0125.

Maybe I misunderstood the forum but it seems to make sense to me.

The guide states that one should reduce the multiplier for the Super LFM at the lowest voltage - this doesnt seem to make sense to me. Why run at a lower clockspeed if it uses the same voltage anyway? Just means its going to clock up again faster and use more voltage.

What do you guys think?

All the best from Berlin,
-COW


PS: link to forum (German)