The "Undervolting" Guide

I've got a T9400 in a Lenovo T-400. I started following the guide, as soon as I selected Performance on Demand (step 2 of part 3) the multiplier went down to 6x from 9x (speed from 2.53 GHz to 1.68 GHz). I shifted back to no management and noticed that CPU-z was only reporting a 2.39 GHz clock speed now. In RM it displayed that the throttle had dropped down from mirroring the core clock to a level about 114 MHz below.

I'm a little confused....since the only way I can get the throttle back the way it was is to reboot

 

I will be trying to do this soon. Thanks for the guide.

 

Are there any alternatives to Rightmark Clock. RIghtmark is wonderful for undervolting, but it has a pile of additional features that I do not use. Particularly the monitoring components. RM clock uses roughly 3-4% of my CPU when in idle and I would prefer a tool which does nothing but set core voltages. Alternatively is there a way to disable unused functionality in RM Clock?

Thanks in advance

 

Without making use of Google, I know of no program that uses less resources than RMClock to undervolt, but give Notebook Hardware Control a whirl. As for disabling features of RMClock that you do not use, there is no way of doing that without modifying the program and seeing how it is commercial/closed-source, that's probably not an option.

That said, does the CPU usage by RMClock negatively affect your system in any way, other than more being used than you wish it to?

 

hi guys
let me see if i can explain this well.
my processor is a t7350 and i know it uses half-multipliers, but:

1) withs rmclock i only have 6.0x and 7.0x FID, also with CrystalCPUID the FID's without an (X) are 6.0x, 6.5x,
7.0x and 7.5x (just this ones? you guys have 11x)?
2)the VID for this ones at rmclock are from 1.1000V to 1.0000V but at crystalCPUID, the alowed VID's (the ones witout an (X)) are from 1.212V to 1.113V.

i guess crystalCPUID is the one with te right values, right??
so, i can't go lower than 1.113V???

help, and thks

 

Hey, can anyone tell me if this is normal?

Here are my values for my T7500 C2D in a HP dv6500t:

6x-.8500
7x-.8625
8x-.8750
9x-.9250
10x-.9500
11x-.9875

Comparing these to the results other people got, I think these results are pretty good. However, when after running ORTHOS with these setting for only a short time (10 mins) the temperature shoots to about 82 C. That seems to be really bad. Am I missing something?

I recently loaded Windows 7 as my only OS on the laptop, previously I had Vista installed. The hardware requirements are practically the same for Windows 7 and Vista. If anything, windows 7 demands less power.

I just recently changed the 11x value from 1.000 to .9875, the current value. I ran ORTHOS for 6 hours at the 1.000 value a couple months ago, with the temp. peaking at 79, and determined it to be stable. Why is it running hotter now? at a lower voltage? Any help is appreciated, thanks.

 

Hi. Great guide, very easy to follow and quick to understand. From the voltages and temperature drops achieved, it seems like I have a not-too-great chip, but it's still decent enough. I also have one of those freak chips with different temperatures between the cores (they idle at 17°C and 36°C). Here are my "last stable" results:

Asus M50SV-A1
Core 2 Duo, T9300

P-state, Multiplier, Voltage, Temperature (#0 / #1 // THRM / GPU / HDD)
Temps are all after 45min ORTHOS Small FFTs

P0, 6.0x, 0.9500V, -- (SuperLFM)
P1, 6.0x, 0.9500V, --
P2, 7.0x, 0.9500V, --
P3, 8.0x, 0.9500V, --
P4, 9.0x, 0.9500V, --
P5, 10.0x, 0.9500V, 43°C / 57°C // 64°C / 70°C / 35°C
P6, 11.0x, 0.9875V, 44°C / 59°C // 66°C / 70°C / 34°C
P7, 12.0x, 1.0375V, 51°C / 65°C // 71°C / 74°C / 35°C
P8, 13.0x, 1.2250V, -- (IDA)

Stock is 12.5x, 1.5V, 56°C / 67°C // 76°C / 75°C / 38°C (after 10min)


Not too thrilled, especially since I'm losing a bit of performance at the top. Will use only for on-the-go. Anyway, my question builds on one which was already asked before - those lower multipliers at the minimum voltage. Would it make sense to deactivate P-states 1 through 4 since their use will not result in any heat or battery-life gain, hence making the system more responsive?

 

I undervolted to 1.0V from 1.15V and when I set the maximal performance profile the whole laptop will shutdown for no reason. Temps are about 60 degrees C

 

I just did my x-9000 , temps are great , at 100% for 3 hours between 53c and 58c . this is at 1.125v instead of 1.2 stock at idle it is at 30- 33c , very impressed . Temps are as low as my t-7500 , much lower than at stock v , at 100% it hit over 70 c . A 44 w chip running cooler than the 35 watt chip is great . Will try running 1.100 and 1.00 when I have more time
Thanks

 

Hello, I want to undervolt my P8600 but none of the programs I tried (RM Clock, Speedfan, HWMonitor) display the correct temperature for my cpu cores. #1 says 38°C and #2 says 43°C no matter what I do. It doesn't change when I do the stress test in Orthos although both cores are clearly stressed. What can be the problem here? I'm using Vista Ultimate x64.

 

You can also try notebook hardware control, it works in much the same way as cpu rightmark.

To measure core temperatures, try using "coretemp"

Are you by chance using an enginnering sample?
If you are, you will not be able to see the core temperatures.

K-TRON

 

Thanks for the hint, unfortunately, Coretemp doesn't do it correctly either Also, it somethimes happens that HWMonitor doesn't display two cores but just one of them, reading 15°C, which is pretty impossible. After every reboot, the programs display a different temperature, so I guess they display the temperature that was correct right after rebooting, but then they are stuck. Could this be a driver issue?
And no, I don't think I have an engineering example, they wouldn't put one of these into the average consumer's notebook, would they?

 

Do you have it set to use only one core in the bios ? In CPUZ You would have es somewhere in specifcation and it would not list model in specifications of CPU if it were an ES

 

No it's not set to use just one core. Sometimes, HWMonitor displays 2 cores and sometimes just one. However, the temps are always wrong. It reads the GPU and ACPI and HDD temps correctly, though. I hope this is not yet another Vista x64 issue.

 

Have you updated to the latest BIOS? (sounds like it's not reading properly)

 

thanks for the guide lowered my cpu from 1.375 to a stable 0.9875 got the BOD at 0.9625

Mark

 

Thanks for the guide, flipfire.

I've followed most of the advice from it. I started by stressing the CPU at 1.275v, the stock voltage, to check max temp. After 15 minutes I began dropping the voltage down two steps until I found an error in the first 15 minutes. I received an error within 4 minutes at the lowest voltage of 0.9v. I then bumped it up one step to 0.9125v where it's been running for almost 2 hours within any problems.

ACPI - 73c -> 57c = -16c
CPU - 72c -> 56c = -16c
GPU - 77c -> 69c = -8c
HDD - 38 -> 36c = -2c

Those are comparisons of the max temperature reading from HWMonitor while running Orthos at 1.275v and 0.925v, respectively.

I'm pleased with these numbers so far. Loosing every bit of heat from the CPU is good but I'm happy with the lower temperature on the GPU. Hopefully it will help prolong the life of the failure-proned part until I replace this notebook in the next year or two.

I'll be leaving Orthos running overnight soon so I can make sure it's stable but so far so good.

 

I have an old Acer Aspire 1650 and I did the undervolting. When I finished i run the firefox and it crashed. Actually, it crashes each time when i start it and cpu rightmark 2.35 is running in the background parallel. Once i exit from RM, it works again. Everything works when RM is running as well, just FF doesnt for some reason. The voltages are set correctly, tested for days. So it is stabile with RM setting...

any idea of why having that problem?

thanks in advance

anpe008

 

Yep, the BIOS is up to date. Any other ideas? Is there some kind of driver required for reading the CPU temps or what?

 

Hi i have acer aspire 5610 with t2050 intel processor. I lowered voltage to from 1.26v to 0.95v and everything is stable. RMCLOCK does not show me any lower voltages . Is this motherboard problem or what? Is there any solution for this?

 

That's the minimum voltage that Intel locked your CPU at, there is no direct way to achieve lower voltages.

 

This does not seem to work on my T61. I can undervolt it down so much and my laptop won't Blue Screen! I don't know what's wrong. I followed the instructions word from word!

 

Quick Questions:

So the only thing not ticked is 0 correct - which for me is SUPER LFM?
Untick "Auto Adjust intermediate-states VID" --- So we re-check this? I am assuming no.
Do you keep the p-state transition checked right? Doesnt say to remove checks so figure those stay.
Is it ok if the highest multiplier is lower then the second to highest if you dont change them all?
Lastly, it now says 1.70 ghz as opposed to 2.26. Does that mean I lost that much power - from 2.26 to 1.7 all the time? Or is this only for when the laptop gets to 100% cpu?

 

ok i have a dell studio 1537 with a p8400 and i was wondering is it worth it to underclock the centrino 2 processors when there already at a low wattage?
and if i reinstall vista again after i underclocked, is my processor still underclocked or do i have to do it again? just in case if i do mess up pretty bad since i am new to this

 

Great guide flip! I just finished undervolting my dv5t P8600. Would like to just ask if these settings/changes are normal. I think they should be.
9.0x went from 1.250V -> 0.9875V
And multipliers 6.0x - 8.0x all went from 1.250V -> 0.9250V
0.9250V is the lowest voltage I can get from RMClock.

Stress temperatures dropped from 80/81C to 68C .

Pretty dramatic changes. Still checking if it's 100% stable. I've followed the guide step by step.

 

does this also lower idle temperatures? i rarely ever go to the high temps that are shown in the stress test.

 

Thats normal, it means your CPU can tolerate a much lower voltage. Too bad intel has locked it to certain point.

Underclock is different from undervolt. If your temperatures and battery life are already good, i wouldnt really bother undervolting.

You will need to undervolt everytime you reinstall windows. RMclock needs to be active.

Sounds fine to me, just make sure its rock stable.

Not really, mostly max temps.

 

yeah i meant undervolt and ok thx just making sure cuz if i do mess up i was hoping i can just reinstall vista because i was gonna test it first wen i get my laptop tommorow and then reinstall vista to remove all that bloat ware

and how do u know if its stable or not?

 

perfect... i make my x61t cooler and cooler

cpu runs at 0.6GHZ with 0.9000v

 

Please read the first post of this thread, it will explain how to test for stability.

 

following this great guide i installed rmclock on vista 64 and managed to undervolt my T7300 as follows:

10x..........0.9625 V (0.9500 causes Othos error)
9x............0.9250 V (0.9000 causes BSOD)
8x............0.8750 V (0.8500 causes Orthos error)
7x............0.8500 V (0.8500 is lowest shown, all OK)
6x............0.8500 V (0.8500 is lowest shown, all OK)
6x (Super LFM)............0.8500 V (0.8500 is lowest shown, all OK)

I left 11x (in my case IDA) unticked as i didn't quite understand what rmclock would do here, and as i know by checking in Everest i already have IDA enabled.

Well, all above undervolting lowered my max temp from really scorching 90-92 C (!!! yes it was that high, and over that point cpu would underclock in order to cool down a bit, then it followed this cycle all over again) to a superb 71 C on maximum load.

Everything ok, i was really exhilarating when after playing around a little i noticed the following (really annoying) issues:

- when playing whatever media (dvd, avi, mp3 etc. ) the sound will STUTTER once in a while
- one of my fav small games is yetisports3 which does not require much cpu power; but if i'm playing it now while undervolted it STUTTERS as well.

I did the above both rmclock enebled and disabled, and the issues happen ONLY when undervolted.

Consequently, i tried increasing the voltage a bit as follows:

10x..........0.9750
9x............0.9500
8x............0.9000
7x............0.8500 V (left it untouched)
6x............0.8500 V (left it untouched)
6x (Super LFM)............0.8500 V (left it untouched)

Has anyone else experienced this? Google turned up nothing on it. Perhaps someone may help out.

 

You might need to adjust rmclock's settings for what percent cpu load causes the speed to increase and how quickly it makes the transition. It sounds like your CPU is jumping frequencies causing your stutter.

 

Thanks for the awesome guide, but I want to ask something. Why it is like this:



?

 

I realised I do have this stuttering sometimes too, so I changed the Up/Down transition interval from 200ms to 150ms. So far I haven't noticed much difference, but are there any negative sides to this?

 

Not uncommon, RMClock is reporting the actual settings based on what is coded into the CPU, CPUZ is reporting the actual voltage coming from the SMBus voltage monitor. My laptop always runs a little more voltage than rated too, but also the SMBus sensor used by cpu-z isn't ultra precise so it may not be quite that high.

Too short of a transition time can cause instability, but too long will cause pauses when the cpu load shoots up. You might want to adjust the "target CPU usage %" I believe they call it to a lower # as well.

 

Sounds like its not throttling properly to the right clock speed. This sometimes happens to iTunes too.

Set your SLFM to 8x or disabling.

Also try setting the Target CPU usage to a lower % for battery or AC

Any CPU-z higher than v1.41 will show desktop equivalent voltages. Check the guide notes, i noted it down.

Increases instability when throttling. 150ms is fine though.

 

Did anyone undervolt a t5800 2.0gz. I have the dell studio 15 and I need help with this. Even my laptop run cool I would rather have more battery life.

 

So i under volted my p8400. wen i first stress tested it it ran at 61 c but i still wanted to undervolt it for battery life.
i only had 3 multipliers
6.0-1.1000 volts
7.0-1.1125 volts
8.0-1.1250 volts and it keeps cool at 49 c
im pretty sure i could of have gone down more but was too lazy and i was checking the speed of my cpu and it keeps running at 1.7 ghz is that suppose to happen? what ever voltage i use it still runs at 1.7.

 

So I played around with several settings like Flipfire suggested,
changed up/down "transition interval", tweaked "target cpu usage level", and also discovered a few interesting things like in the management section the "p-states transitions method"

since my issue seems related to changing between p-states, i tried choosing the "perform single-step transitions only" and the stuttering seemed less subsequently.

at times the stuttering disappeared altogether and tried as i might, i couldnt reproduce it, rmclock on or off.

beats me... maybe someone else with similar cpu (T7300) if they have found some stable settings in rmclock for this cpu, be so kind and share those.

the only workaround for themoment is to leave rmclock on for most tasks, and whenever stuttering kicks in just turn rmclock off (or choose no management profile)

As a sidenote, a detailed guide with all features (advanced or not) of rmclock explained in details would be most useful. couldnt find any on developer's site.

Thanks for all help

 

Hey I think i've found it!

I'm 99.99% sure, all of a sudden it struck me when i closed everest!

So over the last few days while ive been installing / tweaking rmclock etc. i alwayws had everest turned on at the OVERCLOCK section so i could constantly check all those parameters besides rmclock.

What do you know? it seems like some sort of conflict is going on , but ONLY when the overclock page in everest is open, none other as far as it seems.

as soon as i closed everest the issue has gone!!!

maybe someone alse can confirm?

 

Do certain laptops or cpus show different voltages? My Studio 15 only goes down to 1.11625v.....there are no other options?

 

Thank you for the great guide. I managed to drop my temps from 100 degrees to around 82-83 degrees. I have an ASUS PRO50V Notebook with an Intel T2130 CPU ( 1.8 GHz , 14x multiplier ) and the cooling was a real problem. I dropped the voltages from 1.2125 to 0.9750 for all the multipliers and with no stability issues when running both games and Orthos's CPU tests.

 

I was just wondering, is a blue screen the only telltale sign that you've undervolted too much? How about the notebook just hard rebooting without a blue screen. This happens to me sometimes when I'm gaming (while gpu OCed) and I was wondering if my undervolt was too much, or is it just my gpu voltage not being high enough? Does increasing cpu voltage (lessening the undervolt) affect the gpu voltage in any way?

 

Great Guide!

 

With my processor after 45 minutes of stress my max temp is 57C. Is that good?

Also, I forgot to set Performance On Demand on startup so I rebooted and it did not start with the undervolt settings. So then I set it to startup. Is this fine? Do I have to uninstall and re-install or no? Everything is working.

Also, everytime I boot up UAC asks me to allow this program. How to I set this to automaticly allow for this program?

 

There is usually a lowest voltage for any option, and it does differ by processor. Most often it's 0.95V on mobile processors. 1.11625 seems very high though. Are you sure you remembered to select Mobile CPU under Advanced CPU Settings? I don't know the minimum for all the processors, but posting your processor might allow someone else to say what the minimum should be.

Nope, hard rebooting is another sign that you've probably undervolted too much. Undervolting the CPU won't affect the GPU voltage - and if your GPU had too low of voltage, I think you'd actually just get a blank screen, not a restart. My guess is the game(s) that are resulting in this are simply using a lot of CPU power, and perhaps the situation is furthered by the increased heat when the GPU is being used at overclocked settings. But it's not the GPU voltage being chaged - to change that you would have to modify your video card's BIOS.

Try increasing the voltage by a setting or two in RMClock, and the sudden reboots should go away.

57C is very good. There's nothing at all to worry about with that temperature.

Can't help you with UAC, sorry. Always had it off when I ran Vista and now I run XP.

 

http://www.amindlost.co.uk/?p=276

The second half of the article explains how to run any program with Administrator privileges at startup, using Task Scheduler.

Good luck

 

Worked like a charm! Thanks buddy!

 

Should I enable IDA or whatever it is called?

I also notice since I undervolted that my fan is on a lot. Is this normal or am I just being picky?

 

Keep IDA disabled, its pretty useless. Doesnt even engage properly.

Ignore the fans, just watch the temps. Now the temps are different, it might be closer to the fan algorithms switch on point.

No they both have separate power sources and separate logic gates. If the voltage is too low it will either BSOD, freeze the screen or give an error.

If its shutting off or hard rebooting, your gpu might be overheating or something. Its also possible the blue screen didnt show up fast enough and just rebooted.