The "Undervolting" Guide

rtrdogs...
I have this same problem for a months now, have posted many questions to this forum and no one has been able to help. I'm using a 1.66ghz AMD on an Hp laptop, i get BSOD when i use RMclock for 30 mins-4hours, even when all the volts are the same as the stock settings.... the strange thing is that when i'm stressing the laptop i never (overnight) get BSOD, but when i'm not doing anything (maybe browsing) and orthos is off then after sometime i get the BSOD. My computer has never been to repairs and apart from this, its been a good lappie. Also, i tried about 2 versions of RMclock and still have the same problem.
Can someone here PLease help me out???

 

are you sure you havent set the 6x-11x multiplier too low? ORTHOS will only test the highest multiplier which is only used on full cpu load.

Minimal cpu load (eg surfing net, listening to music) will run between 6x-9x multiplier. If these are set too low then it will BSOD

Do the "other multiplier stability test" in this guide

 

Here are my settings on a intel t7700:
SuperLFM - 0,850V (multiplier at 8x)
6X - 0,850V
7X - 0,850V
8X - 0,850V
9X - 0,900V
10X - 0,950V
11X - 1,0125V
12X - 1,0625V

Been stable for over 2 weeks now. Under normal usage my cpu temp is around 48-55C. Under CPU stress it goes up to 71-72. Running Vista x64.

 

Hi guys!
I'm "new" I finally got my Dell XPS 1530 with a T9300 Penryn and Im trying to undervolt it. For now it seems to run stable (tested with orthos) with following settings:
6x - 10x -> 0,925V
11x -> 0,95V
12x -> 1,0V
But im still testing the 10x-12x because i want them to be at least 12hrs stable under orthos.

But i still got some questions:
-Does the CPU produce more heat running with 10x and 0,925V than running with 6x and 0,925V? It seems so... but why? The Voltage is the same.
-Why did Intel set the lowest Voltage that high? It seems that the CPU could run on even lower Voltages...
-Is XP able to handle a Penryn like Vista? What disadvantages concerning Powermanagement do I have with XP?

Thanks a lot! This thread was the most helpful I found....

 

12 hour is kind of overkill, this might put a strain on your cpu/fan. 3 hours is more than enough

Higher clock speeds, thats why. even if the same voltage is passing through it. Higher multiplier = high clock speed

To make sure its on a safe range... if it was possible to make it any lower i bet everyone will get BSOD's

Yes XP will work fine but Vista was the only OS that could handle multiple cores. XP has a patch now that can utilize multi-cores aswell. XP will probably have better battery life
-

Can you guys also post your max temps before and after UV'ing? I just wanna see how much it can take off. UV'ing seems to be the most effective for Merom CPU's. Penryns were already built to run cool.

 

Ok, thanks a lot for your help!
Got some more questions...
-Should I set the SLFM on 6x to increase powersaving? It is set on 8x now. Is it right, that this mode is only active when the system is idle? So 6x wold be more effective...

-How do i configure "the rest" to make the powersaving more effective?

-Can I undervolt my GPU in 2D mode??

Thanks guys

 

8x SuperLFM will idle at 1.2ghz'ish. 6x SuperLFM will idle at about 600mhz'ish depending on your CPU. So Yes set it on 6x so it idles at a very low clock speed. Look at the CPU Info page and compare 6x and 8x, see the difference on clock speeds?

your kidding right? this guide is only for the CPU. I dont think you can undervolt a GPU, i know you can underclock it.

 

okok, sorry.. i read something about a programm called nibitor to undervolt a gpu. google. this would be a great way to save some power too...
thanks so far. ill try to configure my rmclock properly and post the results.

 

I used RMclock to force my CPU to run at the minimum FID (4x) then stressed with Orthos for more then 1 hour and nothing happened. I did the same for all the other FIDs (5x, 6x, 7x, 8x) one at a time still nothing happens. My computer seems to freeze in about 30mins-3hours as long as i have RMclock running and i'm doing almost nothing. I've used several versions of RMclock and still have the same problem. Do you think ACPI or some advance setting will have something to do with this??
Will be very grateful if someone can help me fix this problem. Flipfire, do you have any idea??

 

Nibitor is Nvidias BIOS editor, this can be risky stuff if you dont know what your doing. I suggest you stay away from this unless you want your notebook as a paperweight.

I can see that your trying to achieve the best battery life as possible. Is there something wrong with the M1530's battery life?

This guide is probably the best solution for more battery life without compromise. The profile that we setup in this guide is Dynamic, meaning theres no need to make other profiles.

 

Sorry, I can only guess its a bug in RMclock itself. Ill try to figure it out though..

rtrdogs undervolt was perfectly fine until he sent it for repairs to HP due to the frail LCD cables. After he recieved it back, RMclock was causing issues and got the same symptoms/problems as you which really stumps me. The only variable between now and then was HP touched the motherboard/internals

 

i used to work in office with the lowest possible settings without dynamic switching. i just switch to dynamic when i need more power while working. so im working with 2 profiles.

No the battery is fine. i just want my system as silent and as cool as possible.

 

On a T7300 (Asus G1S-X1) I have been running the CPU at these settings just fine for the past two days:

6x - .9000V
7x - .9250V
8x - .9750V
9x - 1.0000V
10x (max) - 1.0250V

The 10x multiplier was at 1.2500V to begin with. It lowered my max CPU temps on both cores from 78 to 69 and my battery life has increased from 2:30 to 2:50 on the Power Saving setting (locked at 6x, .9000V).

Great guide. Thank you.

 

On my T7300 I have..

6x-.8500v
7x-.8500v
8x-.8500v
9x-.8875v
10x-.9375v
all running stable

The 10x multiplier was originally 1.1750v
Is this right? I mean it seems so low..

Also, how can I get it to idle at 800mhz? The lowest it goes is 1.2ghz unless I set it to no management

 

Yeah they do seem low which makes it a very good undervolt. Run it for a few days, if you get no BSOD then id be very very happy if those low voltages are truly stable. If you do get BSOD's then raise the voltages a little higher. Dont be worried about your low voltages. If it passes a 1hr stability test with orthos, then its most likely stable. If your really worried about it then raise them up by .025v

Again, all processors have different voltages tolerances, they are not built equally so dont be suprised if voltages arent the same as others with the same processor

Do you have SuperLFM? If you do, just set it on the 6x multiplier instead of 8x

How much cooler temps did you achieve with undervolting? Please state max temps before and after UV'ing

 

I'll take some pics of the results later when I have time
Yeah, I already have SuperLFM set to 6.0x but it's still 1.20 ghz..

edit got the pics
before.. usually around 73-75


after, hovers at around 66 most of the time


The thing I appreciate the most is the fan noise (lack of ). Before undervolting, the fan would turn on low after 30 seconds and then ramp up to medium in 3 min, and finally to max at around 4.5 min. After undervolting, the fan doesn't go on at all until 4 min, and then ramps up to medium after 7.5 minutes. It never went up to max speed. I wish I could go lower than .8500v..

Thing is, I still can't get it to idle at 800mhz, even with SuperLFM set to 6x. It's stuck at 1.2ghz

 

Your using RMclock cpu page and not CPU-z to measure the SuperLFM right?

Have you made sure that Superflm index is ticked in the Performance on demand Sub profile page??

 

On a T7300 (HP 8510p)...

Before uv'ing i was getting cpu overheat detected warning from RMClock and temps were around 99C...Seem way too high.

So ive followed the steps, set my 10x multiplier which is the highest before IDA one.

default was 1.200. Its been testing for 25 min at .9875 - temps are hovering around 92-94. Lower temps but still crazy hot....

One weird thing i have everest open as well monitoring temps which matches the temps in rmclock, but under voltage values it shows the cpu core @ 1.1v???

 

I think there might be something wrong with your fan or your heatsink isnt seated properly. No processor should run that hot.

Im not sure why everest is showing that voltage. Try using CPU-z, its more accurate

 

I wouldn't be surprised if they change the entire motherboard....they switch those things so much with HP laptops having problems...They must think its the cure for all problems which it is not...

 

Should i not be crashing or the computer shutting down?, ive set renderings to run overnight before (equivilant to stress testing) and ive never had a crash/blue screen on this system (vista 32)

I'll open it up and have a look inside later today...

edit: So ive opened it up and there was a good 1/4" of dust blocking the exhaust...i think i found my problem

 

:facepalm:
I only set it to superflm in the battery not the ac power.. it's now idling at 600mhz

thanks

 

i just undervolted successfully (thanks for the awesome tutorial). is it possible to have RightMark CPU Clock Utility run at startup? i read that the voltage changes only work when RightMark is running and chances are i'll forget to start it up most of the time lol.

 

yeah, its in one of the options.

 

EDIT: I figured it out. I have to set it in the P-state transitions (PST) in the "Performance on demand". What is the difference between this P-state settings in "Performance on demand" and the one you set in the Profiles (without entering any of the other sub-settings?

How are able to get 600MHz? I just enabled SuperLFM, and the lowest it'll go is 800MHz even when I forced the index 0 to 6x and 7x.

See attached pictures.

 

Its in the guide, read near the end. Very easy to do.

If you dont tick the P-State transitions in subprofile page, you wont be able to use it to undervolt in the Main profile page. Its a handy way of setting up different profiles, which seperates it from the Main profile page

REMEMBER: Everytime you untick something on the main profile page, it also unticks it on the subprofile page. So you have to re-tick the subprofile page again.

I never added in the guide to enable SuperFLM. I omitted it from the main guide because some people dont have it and could cause confusion. Ill add it in now

 

T5270 undervolted from 1.275v to .8625v .4125v unvervolt!!! this has got to be some sort of record!

 

If it passes the 2hour stress test then yeah that is a record. Your running 1/3 less than your cpu's standard voltage, which means very good battery life

How long have you had that setting for? Also how much temps did .412v cool off?

 

currently running orthos, seems totally stable. temps went from 77deg to 51deg though it's sitting on a couch cushion which is terrible for cooling I know, also note that I'm not using HWmonitor, the temps listed are those that the proc stabilized at after 30 minutes of orthos. I expected the T5270 to be a good undervolter as it's max clockspeed is only 1.4 ghz but this is just ridiculous. I only just tried undervolting it today, it's my roommates computer (though I pretty much held his hand through finding and buying it). The lowest setting is .85v though I'm a little afraid of trying it (I haven't tried it at all).

 

Well a 1.6ghz shouldnt have 1.250v as default to begin with. Thats way too much voltage for 1.6ghz. Mine is 2.2ghz and my default is 1.250v

77 to 51 thats like 26c lost from undervolting. Awesome, a 1.6ghz shouldnt run that hot anyway. You must have gotten a lucky cpu chip

Run all your programs/games, what you usually use. See how it runs for a few days to confirm its fully stable

 

my T5550 was defaulted to 1.25 before i undervolted. 1.83 gHz. now its at 1.0250 running no hotter than 53C when playing COD4 multiplayer. 45C idle

 

yeah it's hard to do that cause this is my roommates lappy and he has zero intensive programs, I watched the "X-Men" DVD while orthos was running, I ran orthos for two hours and have done a virus scan and converted songs with itunes. those are the most hardcore progs installed lol.

I think I did get a lucky cpu.

 

great , even running at 100% my temp is 68 deg , used to be 75 .

 

I have the X9000, and in its normal non-overclocked mode, it has all the voltages for all the multipliers set to 1.2000V. Doesnt this mean that the x6 multiplier should be able to go really much lower then that?
Another thing is that for each multiplier it only lets me lower the voltage to 1.000V at the lowest. Is this normal?

 

this is after hitting the defaults button? damn....

 

You default voltages probably didnt get detected properly, maybe because its an extreme processor and its not fully supported by RMclock. Try clicking the default button again and maybe restarting RMclock

You can try googling what other X9000 users got for undervolting and use those voltages as reference to what combinations are possible. I suggest putting a higher voltage margin if your going to copy someone elses X9000 voltage setup

1.000v as your lowest is normal, you do have a overpowered x9000

 

I disagree, I see no reason why the x9000 shouldn't be able to run at the same voltages for a given multiplier as other lesser penryns. My T8100 and T8300 ran at .95v at 800mhz I'm sure an X9000 could run at .95v up to 2.0ghz it's only at higher clockspeeds that increased voltage is needed. I think you may be able to lower the min. available by changing the 800mhz voltage on the cpu settings page (I think that's the page, I'm writing off the top of my head).

 

I'm having quite a bit of trouble with this.. I've got an xps m1530 with a T7500 and I followed the instructions but I'm getting some unusual behavior. For one, Orthos stresses my core temps to 100C on the default multipliers. You only managed to get 79C! Also, after I click on the 'defaults' button in RMClock, my 11x multiplier is set at 1.2V and 10x is higher, at 1.375V.. that doesn't seem to make much sense to me.. but the other multipliers from 10x steadily decrease.

Then when I actually did the test (I changed the 11x from 1.2V to 1.1V) I got a "cpu core overheat" message after a couple minutes and my computer shutdown, I then repeated at 1.15V, same problem.

any ideas?

 

Thats what i would think too, but in reality... the better the processor the higher the default voltages are. It should be the same voltage for the same clock speeds but its not. Ive compared my T2500, T5600 and T7500 standard voltages using CPU-z only

Are you sure your fans arent clogged with dust? Its either that or your heatsink isnt seated properly. No CPU should heat up that high

Your default voltages were detected incorrectly but since we have the same processor i can give you the default voltages for the T7500. Copy these:

http://img379.imageshack.us/img379/701/step4ee1.jpg

Are you sure you didnt miss a step?

 

i have also got a m1530 with a T9300. It seems that the "cooling results" often vary. maybe the headspread isnt attached correctly... you should check that and contact dell.

 

roger, Hey have you tried resetting your SLFM voltage under "management"? will it decrease your min. voltage available in the drop-down lists?

 

Yup, I followed all the instructions as you listed. I'm basically on idle right now and HWMonitor is reporting 63C, maybe I'm having problems with the temperature being detected properly?

But it also might be as you said, maybe Dell goofed on the heatsink installation, if I open it up and check, what should I be looking for?

Also, is voltage being detected incorrectly a common issue?

EDIT: Ok, something is definitely wrong here, I'm getting around a 10-20C discrepancy between the readings from HWMonitor and Speedfan, RMclock seems to be showing the same (higher) readings as HWMonitor. Regardless, these temps are still way too high, I'm thinking its a hardware issue.

 

Does RMclock and HWMonitor show the same temps?

Try using Coretemp or different cpu temp measuring program. There has been one user here that had inconsistent temps between programs. Check page 2-3

Voltage being detected incorrectly seems to be a common issue. It works for most though. I didnt have a problem getting my T7500 default voltages detected.

I dont suggest opening up your notebook but you should really investigate to why you have high temps first (if temp programs are found accurate)

I dont have RMclock here at the moment but SLFM as in SuperLFM? I dont think your T5270 processor support this. If your voltages were detected properly then thats the lowest it will go

 

RMclock, HWMonitor and Coretemp are all showing the same thing. And right now it looks like it's almost 67C on idle. Is there anything else I can do on the software side to figure out why I have these high temps? I'm not sure if its the fans that are blocked.. I don't think they should be, it's only been around 3 months since I got this laptop.

 

Speedfan isnt accurate, alot of people was fooled into having cooler CPU's when it was actually showing the temps of something else

What are your GPU temps? GPU and CPU share the same heatsink/fan

There shouldnt be accumulated dust in just 3 months. It would take at 6months at least before dust starts to get thick

 

Depending on notebooks its can in 3 months...mine certainly does. It also depends on the environment.

 

Yep, GPU temps are also quite high, almost 75C and im not even doing anything application intensive. I'm guessing it has to be the heatsink now, or maybe its some application in the background? Don't know how it could be though since my cpu usage is quite low.

 

my room is pretty dusty, i only vacuum like every 3 months. Just by looking at my notebook cooler, which i havent cleaned for about 3months aswell. There hasnt been much of a significant build up. Keep in mind my notebook cooler has 3 fans and on alot more than the actual laptop

Paperkut, you might wanna consult the dell section of this forum as i dont know anything about the dell's internal parts. Theres something wrong hardware-wise. Id first check the fan if its working properly. Maybe even try update the BIOS

 

Ok I'll do that. My BIOS is up to date so it's gotta be something else. Thanks for the help thus far!

 

yup...that was me, but my issue was that HWmonitor was showing different temps than RMclock or coretemp. I wound up downgrading from 1.08 HWmonitor to 1.07 and then it started showing the same as coretemp and RMclock. No issues whatsoever since then.