The "Undervolting" Guide

The lowest voltage is different for each CPU model. My T2370 has lowest permitted voltage 0.950v and my T7500 - 0.850v

 

Exactly.
My T2300 goes only as low as 0.95V, but my T7500 goes down to 0.85V

 

Lowest voltage is also different for the same CPU for different people. Flipfire and I have the same T7500 but his goes .8500,v, mine only goes to .9000v.

 

I'm starting to think it's the mb/bios if it's the same cpu.

 

HWmoniter keeps crashing on my machine, is there any alternative programs.

 

Try Coretemp.

 

i tried it. I do not have a dual core, and it says P 4 is not supported.

 

For those that remember, I'm still having a problem with RM Clock keeping its settings on shutdown/reboot. Anytime I start my system, I have to manually start up RMC, reset all the voltages, re-select to performance on demand, etc. This is even after applying settings to keep all that saved and telling it to start via both registery and startup folder.

I tried to re-install it, but to no avail. This is on my main system, but it works flawlessly on my old one. Any other suggestions to fix this?

 

Seems like you're having trouble writing to the registry. Make sure your account is an administrator account. You can also try logging in with the default administrator account(Administrator), then try setting up RMClock again.

 

By the way, why do you guys called it "install" with RMClock? There is no installation actually, just unzip it into any folder you like and run it.

 

any tools to do this on p4's and not dual's.

 

@kevinscotland How about speedfan work for me on Pentium4 and Pentium M.

 

Administrator is the only account I have on there since I'm the only one that uses it.

 

@Mooncatt Did you tried both options for startup 'Start up registry key' and start up folder? You could try setting it manually. Go to Start->Programs, find Start Up, then rightclick->'Explore'. A folder will pop up. Copy RMclock shortcut to this folder.

 

Before you do that i suggest you wipe out the registry first.

Go to the RMclock folder and run RMclock_WipeOut.reg to clean the registry settings

 

Thanks flip for that heads up. Now that you mention it, I remember you talking about the reg wipe in an earlier post but forgot about it. Will do that next time I'm on that system and report how it goes.

 

Also just to make sure, please go to the 'Advanced CPU settings' tab and on the bottom make sure "Apply these settings at startup' box is ticked

Btw guys, what revision cpu's do you have? that might be responsible for the inconsistent lowest voltages. Check the "cpu info" tab

JLbrightbill mines a Merom G0 6FB

 

By the way, do you have any antivirus or antispyware software running in the background that prevents writing to the registry? For example, I'm using BitDefender 2008, it has a registry protection. After RMClock is running and made all the changes and startup setting for the first time, when I reboot my system, while exiting, BD gave me a popup saying RMClock is trying to write to registry, so I just clicked ok and made RMClock in the exception list so that next time BD won't try to block it again.

 

Mine is Merom M0 6FD.

 

T7500 Merom Revision E1 Signature 6FA

 

RM CPU clock doesn't work the way it is shown on the screenshots here. weird, i have p4 and xp.

 

Show us how it looks with the P4. Is that a desktop P4 or a mobile P4M? Either way, speedstep will be limited or maybe non-existent.

John

 

mobile p4. so if it is non-existent are you saying that i am unable to undervolt it.

 

My brother's Pentium 4 2.8Ghz can't be undervolted..too bad coz it's really hot!

 

1. Does it show more than one speed option on the profile page?

2. Does RMClock provide any voltage options? If not, it's stuck at one voltage.

The P4M I tested last year didn't provide any options (see the attached). The Pentium 4 family were best known for their heating ability.

John

 

I though pentium 4 M and D's had the EIST (enhanced speedstep) cpu instruction

If no voltages show, then you cant undervolt. You can try using NHC but it will probably give the same results

 

ah well my temperature is constantly at 75c, with HDD at 35-40.

alienware machine suck for heat.

 

Whoa, is that even usable on your lap?

And does everyone keep LFM turned on even when plugged in? I swear my machine seems slower, at just normal tasks. I turned LFM down from 8x to 6x (like suggested). But it seems like, LFM does not switch to other multipliers as often as I would like to see. Is this what LFM does?

 

i would not suggest anyone use an alienware laptop on there lap, unless they indifferent to the possibility of having babies one day.

 

I didn't see a decrease in performance, the clock speed will go up when it needs to.

Some people actually would want that... :laugh:

 

Hmmm, well I was just extrating some files. And it did "need" to jump up in multiplier and it didn't.

Ahhh well, its fine. I am doing a format here in about a hour anyways.

 

Try set it to 8x which is 800mhz to see if it still does it

I didnt notice any performance difference either

 

So My current Undervolt for my CPU goes as follows
T7200
6 .95
7 .95
8 .9625
9 .9625
10 .975
11 .9875
12 1.025

 

flipfire,

Merom G0 6FB also. That's weird...

I'll get some CPU-Z screenies and post those up. You and eleron have the same bottom voltages, as does every other T7500 I've seen in this thread. I even unlocked all the VIDs in the registry but .8500 or anything lower than .9000v for that matter simply won't stick.

 

Hey this seems weird I am so far able to set every multiplier except the max at the lowest voltage does this seem OK or I am doing something wrong?

 

your enable to touch the first one since its always uses the lowest voltage.
your pst
is pretty much the same
0 = unchangeable
1
2
3
4
5
6
7
8 unchangeable IDE

so its normal

 

Hey guys,

Anyone has IDA ticked in their RMClock profile? I mean it might be not that useful or rarely come on, but I don't mind having it enabled. For those who have IDA ticked, how did you guys do the CPU stress test(ORTHOS) to force testing only one core at 1 extra multiplier to find out the lowest stable voltage?

John or flipfire, what's your take on this?

 

Okay this is my final Undervolt I think just want to make sure it looks normal and I didn't do anything wrong.

What seems weird is all the multipliers under 12 I was able to set all the voltages to the minimum .9250V

 

Mine BSOD's if i lower the IDA voltage more than .050v of the default voltage (1.275v). Meaning I wont make any power/heat savings if i enable it

If youve done 45min stability tests on each multiplier and found stable then its all good.

No its not weird. Thats just how penryns are. Some people got to the similar results as you. Some even managed to run .925v for all multipliers.

You can enable SuperLFM if you like. It will just make your CPU idle at a lower clockspeed

For the record, how much degrees did undervolting cool off?

 

Before UV: max temp 67C
After UV: max temp 58C

It wasn't quite as drastic as I had hoped but that might just be the new Penryn processors.

I am now running ORTHOS for a long haul I want to run it overnight just be be triple sure I am stable.

 

hmm... interesting... so ORTHOS can actually make use of IDA...

It wont make any power/heat savings if you enable it? I thought IDA doesn't come on very often, if it doesn't, then it should still mostly use the regular multipliers, no?

 

No i meant ill have to use the default 1.275v if i wanna use the IDA. The IDA multiplier uses the 11x multiplier but in a different way.

One way or the other its too buggy for me. Wont even let me drop the voltages. If it works for you then so be it.

 

I think you misunderstood me, I mean the IDA is using the 11x multiplier, and since it will come on very rarely, I don't see any problem keeping the 11x @ default 1.275V, cause it will mostly still use your 10x multiplier. That's what I meant... this is how I understand it, unless I'm missing something...

 

flipfire,

Ok I just tried ticking IDA 11x using default 1.2750V. I think you might be right about the buggy part, because with IDA enabled, now my 10x multiplier will jump to 1.1750V at peak even though the setting is at 1.0125V. It forces the 10x multiplier to use only 0.100V less than the IDA 11x.

For the record, I still don't know if ORTHOS can actually test IDA, because while ORTHOS is running, RMClock is still showing I'm using both cores and also only using the 10x multiplier.

*EDITED*

I just did another test setting the IDA 11x to 1.1125V thinking it would make the 10x run at 1.0125V. Guess what, nope, it will force the 10x to use 1.1125V also.

 

As I have said before, finding evidence of IDA actually working is like searching for obscure sub-atomic particles. Watch for long enough and you may see the multiplier temporarily go to 11x (or whatever is your IDA speed). I have seen it once or twice in the past.

The real test is whether you can get a better result for something like SuperPi (which is single-threaded) if IDA is enabled. Maybe, if you have everything else closed down so that the second core is genuinely idle. I get no good evidence that IDA brings any consistent improvement and, with the problems it sometimes causes to the voltages, I have advocated leaving it disabled.

John

 

Alright John, I've decided to give up my quest of obscure sub-atomic particles for now.

After a few tests, even though showing no error or BSOD yet, but it still forces the 10x multiplier to use either 0.1000V lower than default IDA or whatever voltage you have set if you set something lower than default for the IDA. For now I will keep it disabled. I mean I can always set the IDA to the same voltage as the 10x, but don't think it's a good idea.

 

If you have a little time run some benchmarks such as SuperPi or wPrime (with one thread) for the same conditions both with and without IDA and see if you can see any consistent difference.

John

 

It seems like it's only a few seconds or less difference, so I decided to leave is disabled for the lower CPU voltage.

By the way, ever wonder why Intel set the default idle clock speed so high even when the CPU is SuperLFM capable? I mean they could easily set it at 1000 or 800 MHz, but they didn't. I'm thinking, maybe these high clock speed CPUs don't like to be idling too low just like a F1 engine doesn't like to be running at low rpm?

 

I don't know why Intel kept increasing the minimum CPU speed. Maybe it was a marketing strategy so we would buy the new CPUs with SLFM. My T7300 spends most of its time ambling along at 600MHz with occasional bursts of faster speed.

John

 

I tried searching on this, and didn't find much.

Does anyone else get really high CPU loads @ idle? According to RMClock, just typing this I am @ 56% CPU load, but 5% OS Load. Seems a little strange. Oh what is the reason that the default prioraty for RMClock is "realtime"? Does it really need that?

M1530 T7250