The "Undervolting" Guide

Looks like this may do it for you. (not personally tried)
http://blogs.techrepublic.com.com/window-on-windows/?p=730

 

Does this guide apply for desktops as well? I'm fairly new to this.

 

Cmon cant anyone help me? Im suffering from constant overheating!

 

Sure but desktops have more cooling options and usually don't need to be underclocked. An enthusiast board BIOS has user controlled voltage. But I suppose there could be a use for RMClock on a desktop ... I can't think of one off hand.

 

Just start RMClock and see if your voltages are listed. If they are then try to adjust them. If not listed, I think you are out of luck.

 

If your motherboard supports it then you can lower the vCore through the BIOS. It will be much simpler doing a BIOS level undervolt.

Sorry i have missed your post. There reason its only showing ACPI thermal zone temperatures is because Pentium 4's never had built-in on-die sensors.

But yes it will work fine for Pentium 4M's

Have you tried cleaning out your fans aswell?

 

I though of suggesting fan cleaning too, good point. And while cleaning you will probably uses a can of air. It's good to hold the fan from spinning too fast to keep from burning out the bearings. A cue tip damp with alcohol or even dry works well, too. Blow the vents out and any HS fins.

 

Post your machine specs, is it a laptop. Maybe you need first to change the thermal compound. To my understanding Pentium 4M only have 2 mode of underclocking.
I used to have a T30 hence my nickname and after applying Arctic Silver 5 I saw a considerably reduction on the temperatures.

Provide us more info on your laptop. If you're concerned with overheating consider upgrading to a newer centrino processor which allows undervolting every multiplier.

 

After completing the undervolt of my laptop I thought I'd post some before, after and idle temps. The high temp. was only decreased by 6 degrees, but I still feel this is more than acceptable considering this was under test conditions (ie before test only ran orthos for 14 minutes, while the after test ran for over 3 hours). Also, the idle temps dropped considerably, prior to undervolting they hovered around the 55-60 degree mark, while now they idle in the 44-48 degree range. Thanx again for the great info flipfire!

 

Okay here are some screen shots of what I've been doing...just wondering if I'm on the right track and doing things correctly?









I know multiplier 12x isn't checked, it's because I'm running the stress test on 11x right now...going to check it again if the stress tests runs fine for an hour or two.

Is there anything more I should be doing with this program that hasn't been posted in this guide? I haven't had the time to browse around RightMark yet...

Sorry, took these quotes from another thread so we both don't have to reply to two separate threads...so is this a problem with my processor? And is this a significant problem?

 

Looks fine to me.

Are you still getting core 1/2 temperature inconsistencies?

 

They are constantly 10 degrees apart from eachother, Core 0 being cooler than Core 1. I have a screeny I just took of HWMonitor in my previous post.

 

Well Im getting a new laptop but I need this to last another month till that comes, and then it goes to a friend so the cheapest way to fix the problem would be best for me. Here are some specs. As for cleaning the fan can you guys give some more detailed instructions? I have never really done any of that.


Here you go, from dxdiag, if anything else Ill let you know.
And yeah it is a laptop

Microsoft Windows XP Professional
HP Compaq NX9600
Intel Pentium 4 CPU 3.20GHz (2 CPUs)
510MB RAM
Page File: 466MB used, 778MB available
DirectX 9.0c
ATI Mobility Radeon X300 64MB
1680 x 1050 (32 bit)

Another thing
I tried running ORTHOS CPU Loader, but lasted maybe all of 5 seconds before comp overheated and shut down..

 

I just did a clean install with vista x64 so I'm undervolting again..
when I run orthos I'm locked at 1.2ghz. RMClock says 6.0x at .9375v but it should be 10.0x at .9375v. ideas?

 

Hi all.

Thx to flipfire (what was that comedy line: "Gee, I hope you didn't"), I now run my T5750 mobile at 0.950 volts for both 1 and 2 GHz (6X, 12X). I don't even bother with the intermediates. When I step on the gas, I want everything, right now, for the application I started. When I'm done, t's nap time. Perfect. Tmax is now a liveable 67C instead of 84C. Apparently, some people aren't aware that the Intel Mobile 'T' series don't have heat spreaders or heatsinks "because their low 35 watt power consumption" doesn't need the draconian cooling of a 100 watt P4. Yeah, right ! Right now, Arctic Silver will only work if you have two heat-conducting surfaces to mate. Otherwise use it to fill a cavity in your tooth if you didn't listen to mom and brush 3X every day.

 

thank you!
i found the link to start RMClock w/o being prompted by UAC at startup from that link you posted.
i just tried it and it works.

+rep

 

go into your profiles tab and make sure you have all your multipliers checked.

 

No problem ... Thanks.

 

Yep they're all fine

 

What about Performance on Demand sub-profile page? Make sure all the PST boxes are ticked.

If you choose Maximal Performance profile, it will lock your cpu to a chosen multiplier.

If your notebook is shutting off from this test, then it has major cooling problems.

 

They're all checked on performance on demand except for the sixth one, which is IDA

 

check your rep points. You got another while you were away. More for patient diplomacy than for anything technical.

 

Yeah, how can I fix this guys? Maybe I should start cleaning my fans before trying to underwatt.. Ive asked for help with a notebook cooler but im being sent to very expensive options, considering I need this to last for about a month. I dont mind putting in the time and effort and a bit of money to get this working.
Right now just using a Huge fan blowing in from the side to get it kinda running. Anything intensive and it overheats and shuts off. I cant take the fan as its impractical and takes up space so need this to do the job by itself

 

I'm also running a T5750. I got lucky and dropped 12X down to 0.950 just like 6X. Here's the deal: These stock voltages are locked into your factory bios, aka firmware. They're not supposed be adjusted by mere consumers (troglodytes) like you and me. But when mfgrs. sell us expensive notebooks with inadequate cooling, whatta you going to do? You use aftermarket utilities to allow your rig to run cooler, like RMClock. Unless you can reverse-engineer the factory bios chip and program in put your own voltages, forget it and be happy RMClock is around. RMClock can bypass the factory bios only as long as the program is running. Remove it, and you go back to factory specs. Which Brand do have, if you don't mind. I'm trying to put together a list of "Overheaters Anonymous" for flipfire as pmt. for his contributions.

 

I am also able to run my CPU undervolted, with VID at 0.950V and max multiplier.

 

T8300(2.4ghz) on a Dell XPS M1330
untouched, CPU temp maxed out at 69C
undervolted to 12x 1.1125V from 1.1375V, temp maxed out at 65C

dont have enough time to run 45-minute stress tests, will do it soon

this saves battery too right?

EDIT: one more thing, this undervolts the max CPU voltage draw, but gives the same performance..
for idle states, u cant do anything right? (besides SuperFLM...).. .idling shows just under 600mhz, 6.0x 0.95v

 

ohhh nice UV! the lowest I could get at the 14x multiplier is 1.13, and @ the 15x multiplier was 1.2.

 

I've messed up the heat-sink real good (no thermal compound, loads of finger-prints and stuff), and still it is at 44*C.

Lowest/Weakest Mobile Core 2 Duo available
But good voltage tolerance

 

I can't find a good explanation of how the max temp throttling works on a Turion CPU.

3 poorly stated questions.
1 Is this a built in process of the CPU and no driver is needed?
2 Or does it need the AMD processor driver to work?
3 Is this the same for XP and Vista?

Specific to RMClock: Is there a way to undervolt and still preserve the "max temp throttle" function of a Turion?

 

Is this possible?

I have T5750 CPU 2.0GHZ Core 2 Duo.

So far, I've been able to bring the voltages down from 1.25V to a staggering (to me) 1.00V!

I want ensure that I'm doing this right. It seems quite startling that I was able to bring it down this much. I've run Orthos for atleast 50mins without nary a problem.

HOWEVER, you'd think I'd have a significant drop in temperature. And is some ways I did.

After running Orthos for 50mins @ 1.00V the max CPU temp was 62 degrees celcius. At 1.25V after 15mins on orthos it reach 73 degrees celcius. Not bad you say, about 10 degree drop. My problem is that idling temps have remained moreless the same? Why? Before it idled at 45 to 50 degrees, now its between 42 and 48 degrees C.

Why was that drop so minimal?

 

idle.... ur CPU is still idling at the same (or similar) voltage... theres a BIOS/mobo set minimum level u cant go under(think its .925).....

 

Quicky question... Why can't I reduce the voltage to below 1.00V when the default (Without using RM; I checked with CPU-z) voltage is 0.9V...

In effect, using RM is actually RAISING temp, anyone got any ideas?

I'm using a Dell Vostro 1500 with a T5470 @1.6Ghz (Windows XP, SP2)

 

Go to the Advanced CPU settings page in RMclock

Make sure the Mobile button is detected. Click Apply and restart RMclock.

 

Hey flip (or is it the famous Mr. Fire by now) Without testing every one of the Orthos test options, which one of the bunch is the most severe for our purposes. I was lulled into a false sense of security until I got a crash with the "blend" test. And, how does one define a "custom" test on the Orthos menu?

Cheers, mate.

 

By default, i recommended the Blend test which stressed both CPU and RAM. I should really only be recommending a CPU test...

If your RAM has issues, it might also cause an error or BSOD after hours of blend testing. Faulty/corrupt RAM is quite common. I suggest running memtest86+ to make sure your RAM is okay.

I suggest you choose Small FFT's to concentrate on the CPU only. Theres also Intel TAT if you wanna stress your cpu

OC'd Desktops usually get tested 8 hours on Small FFT's and another 8 hours on Large FFT's to ensure all parts are in the motherboard are stable. I only recommend testing the max multiplier for about 4 hours, since we will rarely use that multiplier for 4 hour straight.

EDIT: Okay Ive revised the guide so that users will use Small FFT's instead of a blend test. At least if it errors or BSOD, we know it will be purely caused by the CPU.

 

Hi everyone,

Great guide flipfire. I used it to undervolt my AMD Turion TL-60 2ghz processor last night. However, I'm still playing around with the voltage setting. Right now I have:

4x @ 7.75v
10x @ 1.025v

(I let rmclock auto-adjust the middle settings).

At these setting my CPU temps are:

At idle:
core 1: 51c
core 2: 42c

At full load:
core 1: 70c
core 2: 64c

The idle temps haven't changed at all after undervolting, but are about 8c cooler at full load. So far this has been stable but I haven't been able to stress test for longer than 30 mins (just haven't had time yet).

Is there any way for to get the idle temps a little lower? I feel like mine are higher than most of the numbers I've seen in these forums.

Also, when I was testing the different voltage setting in the beginning, I never once got a BSOD. What did happen every time was that the laptop would turn off and restarted by itself. It also happened randomly both while idle and while at partial load. I plan to run more test tonight but does anybody know if this was more indicate of setting 4x too low or setting 10x too low?

Thanks in advance.

 

This guide solved the framerate stuttering issue that I had always thought was linked to my gpu(8600m GT). It turns out that it was my cpu overheating and throttling down to prevent damage. This throttling caused my framerate in games and applications to plummet; regardless of my gpu's clock, temperature, or powermizer settings.

Previous to this guide my cpu's temperature would idle around 63c and skyrocket to 98c under full load. After using rmclock and finding a stable voltage, I was able to drop my idle temps to 45c and load temps to 72c; that's a 26 degree difference while my processor is at 10x multiplier.


C2D T7300

Default

6x 0.8500
7x 0.9250
8x 1.0125
9x 1.0875
10x 1.1625


After rmclock

6x 0.8500
7x 0.8500
8x 0.8625
9x 0.9000
10x 0.9125



There's a drastic disparity between the two settings and I think I could lower the voltages even more, but I haven't had any stability issues so I think i'll stick with these.

 

Nicely done !
Why do the laptop manufactrers keep dumping expensive notebooks on us that are so poorly designed that they can't stand up to the published limits of the CPU makers. For mobile Core-2's, 85C is redline and 100C is immedate, emergency shutdown. I think Intel did their job, printing it in 12 or languages.
How many average notebook consumers know how undervolt? 2%? Geez!

 

Whenever I run RMClock I don't see an advanced tab..what am I doing wrong?

 

I have the same cpu and default is 1.25v

If I understand you correctly, it is supposed to 'jump around' as your cpu changes it's multiplier as it is needed to do more work.

 

T8300:

all multipliers except IDA and 12x: 0.95V
IDA and 12x: 1.0625V

Thanks OP!

 

I finished!

Maximum Multiplier 12.0x: 1.0000V
Everything else: 0.9500V

Thanks for the great guide!

 

Great guide! I got my T5750 down to 1.000v on the maximum multiplier, bringing my max load temps from 78-79C to 61-62C. Unfortunately, this had no benefit towards my GPU overclocking endeavors, but I'll live.

 

Will this work on a Clevo M571RU-U?

 

If all my lower multipliers are on the minimum available voltage already, is there any reason to enable them at all? Why not go for the highest multiplier that functions at the lower voltage level?

Is there any benefit to using less FSB mhz or a lower multiplier if the CPU is getting the same amount of voltage anyhow?

 

even though the voltage is all the same the higher multiplier would probably use a greater current so it would use more power

 

Oh, I see.

Does RMClock have to stay on to work? It takes a considerable amount of CPU time. Can't it just affect the changes and quit?

 

I got my problem fixed, for some reason it locks when i check enable thermal monitor 2.
load temps before were 78c, after is 67c
this is after the 5 min of curing time i gave the as5

Thanks again, flipfire!

 

Mine cannot go under 1.025. Does having a different computer may be the cause?

 

wow, everyones temps are so low compared to mine. for the past year i had my t7200 undervolted to 0.9750v @ 12x and max temp of like 80c with fan at 100%. but recently (more like 2 days ago, gota love hp business support), i had to send my laptop in due to 'heat issues' and hp claims they replaced my cpu, so i went and tested the voltages again, this time i can get it down to 0.9625v @ 12x with max temp of 80c with fan at 80%. still hot compared to everyone elses. i dont even want to test max temp when i dont undervolt, since the max voltage is 1.1875v. (edit, just tested it, and well, it hit 100c and laptop shutdown due to overheat protection)

hum, seem they updated my bios for me as well, maybe that might also have an effect on the undervolt
for those who might be wondering, i use 'coredamage' to heat my cpu up, it gets hotter then orthos does, and i use the acpi values provided by HWMonitor to determine my fan speeds.