The "Undervolting" Guide

currently running 12.0x, 0.9625V with T9300 stress-testing. do we really have to stick with 45mins stress-test?

 

The longer you stress test the more likely you are to find any possible errors that may occur down the road. Generally anything more than 5-10 minutes eliminates gross instability, but the longer you go the more likely your settings are 100% solid. Many computers out there are overclocked or undervolted at settings that may really not be 100% reliable, but are 99.999% reliable and good enough for the owner of the equipment.

 

thanks mike. that's what i thought too. well, i guess i'll just go with 5mins. heh. XD

 

did anyone else have trouble undervolting on a t8300? i was just undervolting my max multiplier and i bsod like 5 times... and its only 7x multiplier? how are people getting 11x as their multiplier? also i have a dell xps1530 if that has anything to do with it...

 

Did somebody resolve the issue with 9.5 multiplier in RMClock with C2D T9400? Or is there any other tool which can manage half stepping? I try CrystalCPUID but I'm not sure about using that.

 

As mentioned in another thread, you could also try CPU Genie (however, it isn't free, $15).

 

For all people who undervolted their T9400. What program did you use? Did anyone use NHC?

 

NHC wont work, try using CPUGenie, RMclock or Crystal CPUID

 

When you lock your CPU to full speed/throttle does RightMark cpu utility need to be active or does this embed itself in the windows power profile so you that you don't need rightmark on.

 

Anyone figure out how to UV a T9400 please post. I can't for the life of me. I can cut clocks and use RMClock but CPUID just crashes my system?

 

Try CPUGenie for 30 days and see if its worth it.

 

can u help me uv my t8300? i undervolted my p8400 perfectly but am having trouble uving it i either get bsod or something crashes and just freezes up and i dont get bsod but i have to turn it off and turn it on again. also i cant get anywhere near what other people have been undervolting my cpu... do u have any idea wats wrong?

edit* or orthos gets stopped...

 

CPUgenie seems to be good tool but I didn't test it yet. I gona test it as soon as I get my T9400 which should be on monday. It is made also 64bit which I need.

 

OK, very happy with the way this went. Flawlessly, actually.

This, on the P8600 in my sig:

Decreased my highest multi VID 0.1375V to 1.0125V; Have all but the second highest multi at the lowest VID available (0.9250V); And am Orthos stable >2hrs at each multi.

Full load temps dropped ~10 -12*C

Oh, SLFM works too, now, giving me an idle of 800MHz.

Final Validation Screenie:

 

I used RightMark cpu utility to lock my cpu to full speed. If i'm on a different power profile ie power saver and I start up Rightmark cpu - I notice that rightmark cpu will change my power profile to maximum performance. However, when you change your power profile on windows with Rightmark cpu on, Rightmark cpu doesn't acknowledge the change of profiles (the red circle still remains locked when i'm not on maximum perf porfile). My question is which one is actually dominate in dictating the computer's power profile when Rightmark Cpu is turned on - the windows power profile manager or the Rightmark cpu utility? It would make sense for Rightmark Cpu's clock to turn gray since I've only set the cpu to lock on full speed under the maximum performance profile and none of the others.

Heres a screenshot of what I mean:
http://img3.imageshack.us/my.php?image=examplet.jpg

Also this leads me to another question - does it matter if Rightmark Cpu utility is on for the cpu full speed lock to take place?

 

I have a p8600 and no matter what I do, CPUZ tells me the CPU only uses two multipliers - 6x @ 1.6ghz and 9x at 2.4 ghz. These are the max and min multis, but it never hops in between.

Anyone know whats up?

 

I also have a P8600 (in sig).

It's likely the CPU is transitioning thru the middle multipliers faster than CPU-Z can poll, or you can see. Your CPU will generally spend very little time, if any, on the middle multis.

BTW, I really like my P8600. It and/or the P8400 seem to be a sort of "sweet spot" in the current Intel mobile line-up, IMHO.

 

Thanks for the reply StratCat.

Your screenshot has inspired me to give undervolting a go pretty soon. Think it's likely I could pull 2.4ghz @ 1.0125V as you have?

Hope so.. I normally idle around 27-32C, max I've seen it has been 46 or 46? All at 9x.

I'm on a t400

 

I've followed the guide to the letter, but as soon as I lower the highest multiplier down by just a smidge and run the stress test, my CPU's temp skyrockets (from ~53 to up to 70s). Is there something I'm doing wrong?

RMClock recognizes the processor as a Core 2 Family (it's a P-6860fx, so it's a 1.83gHz Intel). Should it list it as more specific?

Am I doing something wrong, or am I simply just not able to undervolt?

 

I want to share my results.

I have SZ360P / Intel T7200 Meron (Socket M)

Multipler x 12.0 @ 1.0500 volt.

CPU Temp @ Max Load drop from 80C to 70C with AS5 thermal paste.

 

Have you stress tested at stock before you lowered any voltages, in order to establish a baseline? Just wondering where you're getting your first temp from.


Orthos, Prime, OCCT et al, provide a very significant load. It's common to see an increase of 20*C or more from idle to full load when stress testing.

 

Even with individual CPU variances I don't see why yours shouldn't at least be in the ballpark, if your CPU is a P8400/8600.

Those are very good temp for a stock machine, IMHO. Slightly lower than mine were prior to undervolting.

 

your idle temps were reduced?

 

Not sure 'cuz I only noted my temps when baselining under full load. I couldn't lower my lowest multi 's VID anymore than stock, but my idle speed was cut in half when I used RMClock to invoke VLFM.

Not being able to lower idle VID would seem to imply no temp change. Cutting my idle clockspeed in half via VLFM would seem to imply some possible temp lowering.

I assumed ccol's high temp was load.

My load temps definitely decreased big time.

 

Just a quick question....I have undervolted my laptop to the lowest possible VID which is .7125V (used the tweak) which is great except that this seems really REALLY low as the stock voltages were around 1.1 or something. i dont know if i have done anything wrong or not.

also when i stress the cpu with orthos the cpu info tab in rmclock shows under current...my multiplier as 9.0x which is correct but my "req.Vcore(VID)" as 0.9500V (which was the lowest possible VID before i applied the tweak). does this mean that my current VID for 9.0x is .9500V or does it not display the VID from the tweak?? which VID is correct the one showing on the cpu tab or the VID shown under profiles??

 

Your current VID is .95V, apparently unlocking doesn't do anything unless you have an ES CPU in which case, the voltages might be unlocked and just unavailable until you use the tweak.

 

I should have stated my number properly. At idle, the temperature ranges anywhere from 35 to 50. Upon stressing (without settings), the temperature rises to about a steady 70C. However, after applying the profile in RMClock, after just a few seconds of testing it jumps well over 70C (in just a few seconds it can hit 77C).

 

Has anyone seen some undervolting done on SP series Peryns? What are the lowest voltages?

I managed to set my t5470 to 0.9V on all the supported multipliers(6x - 8X), but still it goes up to 1.175 sometimes when changing frequencies, but only for a moment (See attachment).
It's orthos stable at 0.9v, 1600MHz.

 

today i decided to undervolt my t7250 when i started with highest multiplier and decreased it by 0.025 then started ORTHOS after 10min of test these were the results and i feared that i will fry up my cpu i stopped the test.my question is is it ok to get such high temps i am sure it will go even beyond because these were only after 10min of test.

help me.

 

Wow. Try cleaning up the vents, and also try dropping the voltage for the max multiplier by 0.050-0.100V or more, and then run Orthos.

 

drop voltage for the max multi to 1.0v and run orthos again

 

my vents are clean..will drop the voltage and post the results

 

got BSoD when i dropped it to 1.0v

 

can the vid of 10x multipier be lower than that of the 9x??

 

because run a stability test at 1.1125v for 10x and it was stable so going to reduce it further to 1.0875v but my 9x multiplier has 1.1000v

 

My temps after undervolting:

-full load: 62ºC
-idle: 28ºC

Niiiiiiiice!

 

My t7100 was 1.1750V. Now it's 1.000V running stable even when playing HL2.

 

Changed to 0.9875V and it's running ok!

 

weee... mine dropped 48C to 33C bery goode

 

yes, it can be set like that, and it will work, but you can reduce the voltage for the 9x multi also at least to the level of the 10X multi.

 

one question: is it possible to overclock my cpu? i have a t7100 and i wish to run @ 2ghz. I'm running with .9875v so i thought that maybe i could put @ 2ghz and 1v or a little bit more without compromising the battery life or getting high temps...

 

There's a guide to laptop overclocking here.

The original post is a bit outdated, but the software linked to should work fine.

I can not recommend overclocking your laptop though, as I feel that the risks far outweigh any real benefits.

Good luck.

 

to undervolt while overclocking you have to overclock the processor and stress the processor at that speed then gradually undervolt it from there as opposed to undervolting your stock speed then trying to overclock, as it might crash from lack of power... this is what I got from previous posts anyway.

My attempts at undervolting so far with my t2250 have reduced my temps from 73 degrees to 61 under load and I've already had to get my motherboard replaced due to my GPU burning itself out, so this is really good for my laptop.

 

Read this excellent guide and started playing with RMclock on my E6400 with a P8600. One thing I don't understand: when I use the Dell Power Plan, the core clock is running at 800 MHz or so when idling. If I use the RMClock power plan, and select PST in Performance on Demand, with all the PST states checked, the core clock runs at 1,600MHz when idling. This stays the same even when I lower the core voltages. Does the higher core clock mean that there is more power consumed?

 

sry to ask, at 9.0x p8600 processor 1.0125V ORTHOS comes the error at 2 hours stress testing.

with 1.0250V ORTHOS won't a problem after 5+ hours of testing. is the 1.0250V okay to use or will i use 1.0375V? thank you and god bless.

 

T9300 undervolt results (using CPUGenie)

6.0-11.5x -- 0.950v
12x -- 0.975v
12.5x -- 1.000v

Idle Temp: 42C (original) --> 40C
Max Temp: 67C (original --> 61C

So here's the question: it's a decent improvement (~.100v or more at each multiplier, a few degrees cooler at max load), I guess, but will it make a significant enough difference in battery life to offset the RAM usage (and the $14.99 for CPUGenie after the trial period)?

 

got a BSoD at 1.025v for 10x after 10min of testing should i fix with 1.0375v for 10x or go a little higher ? i tested the 1.0375v for 1 hr and got no problem.

 

I found new beta version of Notebook Hardware Controll... looks VERY NICE.
http://www.notebook-treff.de/board/index.php?page=DownloadDBData&dataID=116.



Actually there is NHC 2.0.7.3 beta from yesterday and it is for Vista 32bit for testing purposes only. It seems that it can manage everything including .5 multipliers. But in this beta works only SpeedStep settings, others (like Graphic card or HDD) are locked.

A you have to register on that forum to download it!

 

go for 1.0375v probably, and test it for longer because my other laptop got an error after an hour 7 minutes with a t7300 at 10x at 1.000v and I reverted back to 1.500v and the temp difference wasn't that huge, 2 degrees i think.

Managed to get my t2250 from 73 down to 55/56 degrees (13x at 0.9750v), still stable after two hours stress, got a BSOD at 0.9500v though. brilliant guide.

oh yeah does the application actually need to be running for the settings to take effect? because I've just told it to start up with windows.

 

Thats great news! I havent checked up on their forum for a couple of weeks. Thanks for finding it! The problem is I havent figured out how to download it yet...LOL