The "Undervolting" Guide

Because your processor will consume energy every time it cycles. The faster it runs, the more power it consumes. Think of it as a car:

Your car consumes gas each time the engine cycles. The amount is controlled by the throttle (gas pedal). So each time the engine cycles, it sucks an amount of gas into the engine relative to the current position of the throttle. If it's running at 1000rpm, it's sucking that much gas in 1000 times per minute. At 4000rpm, it's sucking that same amount of gas in (with equal throttle), but 4000 times per minute.

Likewise, your CPU is using power each time it cycles, and bleeding that power off as heat. Lower speed means less cycles and thus less power consumed - and less power converted into heat.

So, a lower SLFM multiple helps conserve battery power and reduce heat while your CPU is at idle.

Edit: As to clocking up faster - that's really a minor difference. The point of SLFM is to hold power drain and heat generation at an absolute minimum when your processor is idle. When you're just doing random web browsing/email, that's the majority of the time. In all likelyhood, it'll ramp up regardless of what multiplier you have it set to when it needs to do something. It's very rare for your processor to only step up a couple multipliers and sit there (though that does happen from time to time). That's also why the intermediate states are not so important to get carefully set (though it doesn't hurt!) - your processor generally only sits on those states for a very, very brief time.

 

Bit of a weird issue here, might be related to my undervolting, might be something else.

I've had blue-screens twice but with unusual effects. See the attached image - the first blue-screen had a graphical "ghosting" of text. The second blue-screen had a red square glitch in the corner. (I've mocked up the image as an example to show both effects as they appeared.)

Question is - can this happen as a result of a normal undervolting crash, or does this indicate I might have a hardware problem with my GPU? I've never seen this sort of thing before.

It doesn't crash during high CPU load, which is odd. Stress testing passes without incident.

 

Can't really tell whether that's related to Uving or not. Best way to check is just bump the voltages up a level or two for a while and see if it goes away, if so you know the voltages were too low.

 

NICEEEEEEEEE!!! Hey, dedicated video cards can be undervolted?

 

What GPU do you have?

SPCMDCON.sys is a system file thats corrupt. (google the filename)

Just as a pre-caution, i suggest turning off UV for now and see if this issue re-appears again

Yes but its not easy at all, you will probably brick your vBIOS. You can only drop the voltage by .5v and only loose 3-5c of heat

 

Cooler is cooler. I still haven't ordered my notebook at Xoticpc (it's still in the shopping cart), do you think the guys there can undervolt my GPU and CPU for me?

PS: I've heard tales about RAM undervolting, what do you think about that?

 

1. They likely won't UV for you. It's not that hard anyways.
2. AFAIK GPU's can't be undervolted since it's predetermined voltages are vBIOS set (however through vBIOS flashing you could modify the upper voltages for more OC room but more than likely brick your GPU)
3. RAM voltage changes AFAIK are only possible on desktop MBs since they allow the vRAM modification. The problem is that I believe they only allow the voltage to be brought up to allow for stable frequencies when OCing.

 

Well, then I'll stick to undervolting the CPU and thus making sure nothing explodes.

 

Any idea why does RMCLock only recognise the highest multiplier for me right?

 

can you show me a screenshot? try turning off your OC

 

Haven't gotten a BSOD yet, but I get errors from ORTHOS, could I maybe still go lower until I get a BSOD?

Also, it says see stress.txt, but I can't find that file in the ORTHOS directory.

 

No thats your limit, dont push it any further. I suggest raising it by a notch and then doing a 3 hour full stability test.

 

Thanks for the quick reply flipfire! I'm worried about doing anything longer than 12 minutes or so with serious heat issues on my HP dv6000 T7200. Even with undervolting of 12x to 1.075V, I'm getting ACPI of 89C and Core temps of 87C

 

Something is really wrong with your notebook, UV'ing wont solve the problem.

Exactly how high were your temps before undervolting?

You might wanna consider sending it in to HP for a checkup.

 

The Cores hit 95C before I stopped ORTHOS before undervolting

 

Holy crap, that is insanely HOT.
You need to do a fan/vent cleanup ASAP.

 

Don't go any lower, in fact, you need to raise it a little like flipfire stated, you're still at risk even with just errors. And your temperature is a little higher than usual.

 

Yeah, I know. I looked at the fan/vents and don't see too much dust. Do you suggest just using compressed air and blowing from the fan towards the vent or taking it apart?

 

Wonderful guide Flipfire Thank you so much.

First off, I am not the kinds who goes around doing these things / following such high tech guides simply because I don't know a damn of these things - way beyond me, to keep it simple. And I am sh** scared to even think of doing it alone, without a guy who knows computers and stuff next to me.

Anyways - I decided to go ahead and try it out simply for the fact that its so wonderfully written

Now to the story. I followed everything. No problems. I had a max of 13x with 1.1375V next to it and as you said, I started from there to undervolt. I wanted to start with ur suggestion of getting it down by 0.100V to begin with and then getting down by 0.025V till we kill the system. But then I thought I will start one notch higher than 0.100V, just to be sure you know...

Peak temp's before doing this was 72/74 in both core', respectively. Then with the undervolt of 1.06.. something, (and not 1.0375 as advised as a starting point in your guide [1.1375 - 0.100V]), temp's peaked around 68-70 in both core'. Then I decided to go with 1.0375, and BANG, the damned blue screen LOL... You have no idea how much that sudden blue screen freaked me out.. and this is the first BSOD in my system ever since I bought it.. LOL.... But the speed at which it happened, combined with the fact that you had said we could go down by 0.100V to being with just didn't make me be prepared.. LOL... :laugh: :laugh: :tongue: :tongue:

Anyways, Then I felt that 5 or 6 C temp changes was not that much a difference making phenomenon and in my case I think I would hardly get my CPU to full load like this orthos stuff.. so I just left it with the default / factory volt values of 1.1375 for the 13x and did not make any changes or whatever...

Anyway... it was fun.. lol... you should have seen my face during the BSOD... lol... I am laughing so much now...

*This post is just to keep you informed in my case and with my system spec, and to say thanks to you for your nice guide. Now I know much more about these things and how they work, than before, thanks to you*

Cheers...

Edit: damned rep limits again!! too much rep' in the last 24 hrs, my foot!!!!!!!!!!

 

ok....i think i'm ready to undervolt...this thread is pretty long, and i don't have the proper thread tools to look up my own posts, so can any1 pls post the link again on how to do this step-by step ?
if there's also a guide on undervolting the GPU's on the M1730, i wanna look into that too....

thanks in advance

 

But the first post of this thread, the OP, the actual guide by flipfire is what you are looking for / asking for, right??? I am just wondering what you mean by step-by-step other than the wonderful first post of this thread!!!

Cheers...

 

Hey flipfire, thanks for the guide, it was really helpful.

Just a quick question, are those voltages normal? I have a T5550 (1.8 Ghz). It seems kinda... low. (I haven't gotten a BSOD yet, and this is the lowest RMClock will let me go).


Temps for reference:


It used to run around 76C (full load) with 1.25V. Thanks!

 

Those voltages are normal, the C2D series can usually go .95V @ 2 GHz like what I'm running now.

 

In the process of undervolting the temperatures between the two cpu cores are different in the HWMonitor. Is this normal cuz all the pics. I have seen the temps are the same. They are usually about 2 to 5 degrees C off of each other. Do I have to worry about this?

 

What CPU/OS? is it an AMD?

Its nothing to worry about

Its possible that one core is just working harder than the other.

 

Vista Ultimate
Core 2 Duo T8300 2.4 GHz

Yeah I kinda figured it wasn't that big of a deal but I just wanted to check.

 

Also, depending on your notebook design, one core might get more cooling than the other.

 

I finished testing multipliers 7-11 on my C2D 7500 2.2, and here they are, all tested for 3h+ (except for the 6x which was on default and SuperLFM which already has the lowest one but I ticked 6x instead of 8x)

Multiplier - now (before)

11x - 1.0000 (1.2125)
10x - 0.9625 (1.1375)
9x - 0.9250 (1.0625)
8x - 0.8750 (1.0000)
7x - 0.8625 (0.9250)
6x - 0.8500 (0.8500)

I never got a BSOD during testing, just MOBO beeps when errors occured, so it ran fine
The weird thing was, I tested my 11x multiplier at 0.9875 for 3h and it ran with no errors, and the next day I just wanted to check all my multipliers for 10min each and the 11x (@ 0.9875) reported an error after a couple of mins of Orthos torture . It just makes no sense, so I raised all my multipliers by one option more just to be on the safe side so the above ones are the current ones I have. It lowered my core temp from 77-83 degrees on max load to about 68-73 (depending on the ambient temp).

This is a great guide!

 

I've tried to RMwipeout.reg but I still can't get my cpu (t7250) to un-underclock. It keeps running at 6.0x (1.2Ghz) when I run orthos.

 

Edit: Fixed it buy running WipeOut.reg and deleting folder contents and placing new program files into folder.

Dang the little clock icon even gets partially red in relation to speed ... cool.

I have a similar thing on a Turion. Performance on Demand previously allowed stepping from 800 to 1600. Now it is stuck on 1600. Not sure what I did.

I did bump up to 4G RAM, but ??? Doubt that would have effect.

I did "uninstall" but same old.

 

On the Performance on Demand profile page, are all the indices checked under P-state transitions for both AC and battery or just the highest one?

 

Just undervolted the Penryn P8400 in my brand new Asus M50Vm-X1. Does RMClock fully support Penryns? I only see the SuperLFM and 3 other multipliers (6x, 7x, 8x), and only 8x, it's only running at 2.12 GHz instead of 2.26 GHz.

 

Its not supposed to support the new montevina platform as far as i know.

can you show me a screenshot of the RMclock pages?

 

Hey guys. I am trying to undervolt my XPS T9300. The lowest voltage i can go is 0.9500. Some ppl with T9300 have the option to go as low as 0.9250.. Is there a fix to this? Or is 0.9500 is the minimum for my system?

Thx!

 

.95V is the minimum for your CPU since Intel hard codes the lowest voltage that the processor can use. Basically, it's luck of the draw my friend.

 

You can go as low as .7125v by going in to the Program Files folder, finding the "tweaks" registry file in the RMClock folder, and running it. Then go to start->run, type "regedit," and then go to edit->find and type in "unlockvid."

When the search finishes, click the unlockvid registry entry and change its value to 1 from 0. Afterwards, restart RMClock and it will let you undervolt down to .7125v (although your CPU most likely won't be stable at this low of a voltage)

 

Im afraid that doesnt work.

 

Hmm, I guess you're right. I was able to unlock the lower voltages in RMClock, but they don't actually apply when I'm using Orthos or Prime95 to test the settings.

Edit:

T8300 is Orthos stable at 0.9500v/2.4 GHz ftw!

 

i was trying to check for current temp before undervolt by running orthos, but it won't start. I got this

 

Looks like something messed up when trying to initialize the program. The only thing you can do is try a different test or delete the program and download a new copy that hopefully isn't corrupted.

 

downloaded new copy, still can not. looks like I'll try different test.

 

Hello all,

Ive successfully undervolted my Thinkpad R61 with the Pentium Dual Core T2370.

Heres the thing; All of the FIDs work at .9500V. ive tested and retested and every one of them is rock stable at .9500V. I think another user, atthegates,
had the same experience.

I know its working because I went from 53c to 43c on idle and 68c to 53c running ORTHOS for a few hours, and the fan rarely comes on any more. (plus I verified with HWMonitor and Coretemp.

So here is my question: If all the FIDs run at the same voltage, is there any reason for me to use an other than 13X? In other words, why should I set the CPU to slow down if its not going to save any energy? I havnt done any scientific testing, but my general impression is that the battery doesnt drain any faster nailed at 13x than 6x, but that might just be wishful thinking?

This guide was so helpful, as I was using NHC for a Pentium M system and was not able to get any of that to work with any new processors!!

 

If I remember correctly, the power is proportional to the CPU clock
and to the square of the voltage. So, if your CPU runs at a lower clock with the same voltage, your lappy still saves some power.

 

Hm well that's interesting lol. Don't have the laptop with me at the moment, I'll get some shots on Friday. Until then, I remember some of the settings available:

SuperLFM 6x to 8x
Normal 6x
Normal 7x
Normal 8x

Min voltage 0.925V, I think max is 1.25V?

 

RMClock doesnt seem to be able to detect which processor i have. Whats up? (Unknown Core?)


I have a T8300

I did

But still no luck

 

Which manufacturer of laptop do you have? We've noticed that some laptops have the CPU properly identified while others don't, this shouldn't make a difference though to your ability to undervolt.

 

im on a gateway M-152XL

so should i go forward with the process anyway?

 

It should work just the same. At worst, your laptop will explode causing death and destruction within a 20 mile radius of the blast site... at best you'll be able to undervolt just like anyone else, RMClock just won't be able to recognize your CPU.

 

Its fine to go ahead Lajkin. That question gets asked quite often.

This issue happens to all the XPS users with Penryns, its not a problem at all.

Voltage doesnt directly mean how much power its drawing.

 

dude sweet stuff here. just undervolted my cpu on my old dv2400t from 1.2375V to 1.1V. temps have dropped to 68C from 79C. speed also dropped from 1.995GHz to 1.994GHz but not a big deal. haven't tested battery life yet though. thanks so much for this guide Flipfire.

i do have some questions though what does this voltage drop translate to in terms of power consumption or how do i determine the power reduction.
once again thanks sooo much for this. +rep.