The "Undervolting" Guide

Just for reference, the rmclock voltage on my T7800 (w/o registry tweak):

SFLM 0.85V
6x 0.85V
7x 0.85V
8x 0.85V
9x 0.875V
10x 0.9125V
11x 0.9625V
12x 1.0125V
13x 1.0625V

Most of the voltages are tested by running orthos for ~6hrs (left it on when sleeping =P )

 

I recall that Intell had locked the voltage on the lowest multiplier in the earlier core 2 duos.
Have they removed this restriction in santa rosa ?

 

whew, just finished finding the optimum clock (maybe).

I found it really strange, I got no BSOD at all, is this intel C2D Merom T5250 something or what? I set up all the multiplier into the lowest voltage (6.0x = 0.9500v, 7.0x= = 0.9500v, 8.0x = 0.9500v, 9.0x = 0.9500v).

Right now I'm running stability test with Othros, and it's running 12 minutes without problem. Hmm, interesting, has anyone else experienced like this? (undervolting without BSOD, and tweaking all the multiplier into lowest voltage without crash)

 

Anadhi: We have had 2 cases now on a T5500 cpu reading incorrect temps with HWmonitor. I think they had to use an older version for it to work

Like said earlier, not all CPU's are built equally. Some can run more efficient that others.

Do 1-2hr stress test on all the multipliers except 6x . After that, run all the programs you normally run to confirm its stability in real time use. If no BSOD, then no problem! If it works.... it works!

lemm4: here you go, too lazy to explain http://en.wikipedia.org/wiki/Dynamic_frequency_scaling . Its a good read actually.

No its still locked, so people arent stupid enough to make their own idle unstable. Even though you can lower it a little. They should only lock this on desktop CPU's as they can UV using the BIOS, and if you set the idle too low, your cpu wont boot up. You need to clear the BIOS settings to get it working again. Notebooks are an exception since we dont UV at boot up, only when windows loads

 

Hmmm, in that case is the OP incorrect ? Reading the OP, I got the impression that one could undervolt all multipliers.

 

I am the OP. Yes you can undervolt all the multipliers except for the lowest multiplier which is the idle multiplier. This is locked on purpose

 

RM Clock on my T7500. Works fine so far, been like this for days. Tested with orthos twice for 3 hours each on different occassions. The moment i reduce the 11.0x to 1.000V, i get BSOD. So yups, I think this is the best I can go.

SuperLFM 6.0x - 0.8500V
Normal 6.0x - 0.8500V
Normal 7.0x - 0.8750V
Normal 8.0x - 0.9000V
Normal 9.0x - 0.9250V
Normal 10.0x - 0.9875V
Normal 11.0x - 1.0125V

what kind of alteration is necessary for IDA? Leave it as it is? I haven actually seen the effects of IDA before. It seems both my cores are always working. hmm.

 

Turn it off, its useless to be honest. Its quite buggy for Merom chips, it causes BSOD's

http://www.theinquirer.net/en/inqui...tel-dumbs-down-dual-santa-rosa-cores-with-ida

They did tests on it and IDA only worked for like 2 seconds

 

hey, can I assume that the procedure outlined in the very first post works on desktop cpu's as well?

I want to try to get my desktop to run cooler during the summer, and also to make sure I dont mess up when i undervolt my laptop.

I have a intel pentium 4, 3.00 Ghz.

Thanks for any feedback.

 

yes, it will work for P4's aswell. The settings just might look a little different but the same procedures still apply

 

flipfire, excuse me if this has been asked in the thread but, can you do the undervolting with a P4? I have a friend with an older laptop that runs pretty hot and I was telling him about the wonders of undervolting but he has a 3gHz P4.

Thanks!

 

It was just asked 2 posts before yours lol

It should be simpler undervolting a desktop CPU, assuming your CPU is supported by RMclock. Heres the list

I believe P4's dont have the EIST (Enhanced Speedstep) technology, so there shouldnt be much multipliers to tweak

 

I've been undervolting for a while but just recently I noticed something that I didn't see before. With any multiplier (6, 8, 10 on my t5450) the voltage seems to fluctuate randomly between .9500V, 1.1000V, and 1.2500V even though they're all set to .9500V. I have checked this out in RMClock's CPU info page as well as a few different versions of CPUz, thus confirming I'm not getting a bad readout from a single program. Anyone else have a problem like this?

On a side note, I'm thinking of getting a T8100 or T8300 (the GM965 supports the 800MHz FSB, correct?) to replace this, so what are the lowest voltages available to each of those chips (I know it can vary slightly) and which would be better for a programmer/3D-modeler?

 

No, are you sure the undervolting is in effect? It doesnt sound like it is. Make sure you ticked all the boxes in the "performance on demand" sub-profile page and on the main profile box, make sure all the drop-down boxes are on Performance on demand then Hit apply.

Are you seeing lowered temps though? Because this matters more than a bunch of inconsistent numbers

Yes it should support it, double check your BIOS supports it and ask around. A T8300 would be your best choice. Penryns are very good already on battery without undervolting so i dont think you should be worried. They are also good UV'ers compared to the previous chips made.

 

It is in effect, what happens though is that on any multiplier, for example, when idling at 6x it will usually stay at .95V but will occasionally spike to 1.1 or 1.25V even though nothing is causing enough of a load to warrant a speed increase.

I am seeing lower temps, my point was that I'm just trying to make it more efficient by cutting out the spikes/wondering if anyone else had something like this happening.

My current BIOS should support it since there have been a few updates to it since it was released/after penryn was released. I was kinda debating about the 300MHz difference and it almost doesn't seem worth it to me for about $50.

 

I am so sorry if I missed it discussed, but has anyone here or know if its possible to undervolt a Intel Core Duo?

I am getting the XPS M1530 - it has the Intel Core 2 Duo 2.4GHZ T8300, and one of the first things I'd love to do with it, is to undervolt it. When I did this on my Inspiron, the temperatures dropped by 10-15, I usually got up to 70C under high performance. Now, (with Arctic Silver), my high is 45C with a constant 31C.

Thanks so much for all your help and of course your time!!!!



PS: When trying out different voltages, should one just go from the lowest, if it freezes just restart and try the next higher voltage? Trial and error?

 

Im not sure whats causing the incosistency in the voltage but the fluctuations are normal, sometimes the system just does so..

Id have to say its worth the extra $50 for the penryn T8300

yes it is, i was able to undervolt my T2500 core duo (yonah). the annoying thing was, it had like 10 multipliers to tweak.

 

What about if I said the prices were $270/$315 for the T8100/T8300 respectively, post tax and expected shipping from the cheapest locations I could find in Canada (damn I hate the price gouging). I guess the T8300 would help future proof it a bit more and might even help my HD2600 in games.

 

Value for money compared to the T8100, the T8300 is still worth it. The T9300 definately isnt, if your buying it retail.

If your really on a tight budget or dont need the extra edge of power then the T8100 will be a better option. Its enough to run all the programs and games these days anyway

 

On my T8300 I've managed to undervolt at 12x 1.0250V, down from a default 1.1375. For the lowest voltage, 0.950V, it seems to have the same temp regardless if I put 6X or 8X @ 0.950V. Should I even bother to let my cpu go down to 6X? Or can I keep it @ 8X 0.950V 800MHz?

 

Your preference, there wont be much temp difference between 600 and 800mhz specially when they are running within the same voltage

Set it at 800mhz if you like

 

Thanks, just for complete 100% clarification, for those that have the XPS M1530, have you successfully been able to change the voltages/undervolt?

And has anyone successfully used NHS/RMClock with their M1530 - not to monitor anything but to succesfully change voltages/settings?

Once again, I have the T8300 CPU. I am just making sure since for those that know DELL, they may have blocked it in their BIOS.

Thanks so much for those that answer this, just need this answer to clear all my confusion and my distractions.

 

Yes, 99% sure you should be able too. I havent heard of manufacturers blocking UV'ing in the BIOS so far. Lots of T8300 have been successfully undervolted. Why dont you just try it for yourself?

If you using XP you can use NHC but for Vista you have to use RMclock

 

I will be able to very soon, haven't yet received it

 

I am successfully undervolting my T8300 on my XPS M1530.

 

Perfect, done deal, thanks that clears that confusion up.

Just for clarification, through which program are you undervolting it?

PS: Whenever you have the time, I'd love if you could share the voltage measurements you set, that be amazing.

THANKS richierich!!!

 

I'm using the best out there: rmclock

So far I've been very happy. Slowly testing each multiplier. Right now stability testing 10x 2.0GHz @ 0.950V. 2 more multipliers to go, and I'll post my settings.

 

Soo with that CPU stress program, if I test it for 20 mins and get 1 error in one of the cores, should I go back to a higher voltage? Or is having 1 error ok?
It just seems that my undervolt is really bad, I cant even go down 0.1000V without getting an error in the stress test

 

So there really is no way to go lower than the capped voltage in rmclock?

 

No because Intel/AMD locks the voltage min. temp, that way there is a safeguard in place.

 

Ok so on my XPS M1530 with T8300 processor, here are my rmclock voltage settings:

6x-10x @ 0.950v
11x @ 0.975v
12x @ 1.0250v

On power saving or performance on demand, my lowest multiplier is 10x since my processor is stable @ 0.950v. Pretty good, I'm glad I paid a little bit more for the T8300.

 

That's very interesting that you kept the same voltage (0.950v) from multiplier 6x to 10x, I thought you were supposed to increase in in increments, but if it works, can't complain.

So, there is not 13x, 14x multiplier or is that not needed?

thanks again

 

Yes I was able to keep the lowest voltage settings for those multiplier settings. You only need to increase voltage if you get BSOD, crashes, etc...On the T8300 the highest multiplier is only 12x.

 

Yep T8300 is the sweet spot for price/performance! good decision, the penryns have *amazing* power per Watt.
you can't go lower at 12x? my T8300 was stable at .975v at 12x (I ran it at 1.0v to be safe though) and some are stable at .95v.

 

I was using T7300 and now T7800. I can use 0.85V for those low multipliers...how come penryns use 0.95V? I thought they have better undervolt ability

 

They do,

Penryns multipliers can all run at like .950v or so.. That beats your low voltage of 0.850v, unless you can run all your multipliers at that low voltage. You are just able to set your min voltages lower but it wont be stable at those voltages. Penryns win because they can set their voltages low all around and still be stable

 

I am experiencing something strange on my Dell XPS M1330 with a T7500. I started out and got a max temp of 72 during the ORTHOS stress test. However, after UVing the 11.ox box in RMClock from 1.3625 to 1.2625 I ran another stress test and experience temps 10 degrees C hotter at 81!

Here is a shot of my RMClock:



Does this have anything to do with the fact that my 10.0x box is higher than my 11.0x?

Thanks for any help.

 

the default voltages were detected incorrectly. Use the default voltages shown in my guide. Its also a T7500.

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

Start undervolting from these voltages

 

It's weird. Rmclock doesn't detect my CPU. It only detects that it is a Core 2 Duo processor. As soon as I lower it past 1.0250v I get the BSOD. Are there any suggestions as to get rmclock to correctly detect my T8300? Maybe that's why I can't lower it? I'm assuming that my chip just isn't as good as yours.

 

well your cpu is a Core 2 Duo. try comparing your BIOS with other XPS users.

From the looks of it, 1.025v is the limit of your processor. Ive had 2 other people with the exact same notebook as mine beat my voltage settings by a mile

be happy with what you got. whats the default 12x voltage anyway?

 

Default 12x voltage is 1.1375. My laptop's BIOS is the latest one from Dell. rmclock doesn't detect my cpu core or revision #. (Penryn?)

 

Is it ok to run this while using the built-in percent system from Asus for how much cpu speed i want?

Right now I have it set to 0-100 for testing, but usually its at 70-100%

Edit: Also, once I edit my max multiplier, I have read that the rest of the voltages are tested, but are all the same? So techincally just find the lowest voltage of the next multiplier and put that for the rest of the normal multiplier boxes?

In cpuz I only see 12x and 6x being used. If I was allowed to use Asus' power manager and I cut down the amount of cpu %, is that where I would see other multipliers? Once stable, I will run rmclock at startup so I can always take advantage of the tweaks.

Thank you for your patience and help.

 

Im not too sure on this, you might wanna ask other users with the same notebook/CPU that had success. Penryns werent fully supported by RMclock. I wish the RMclock team would update the program frequently but they are running way behind

I dont have the ASUS utility but im pretty sure all it does is limit the min/max speed you can reach, which is fine.Though undervolting is mostly beneficial when on full CPU load.

wait... your not talking about the Vista power management are you?

You have to find the stable voltages for each multiplier except for the 6x (idle). Alternately you can just set the voltage slightly lower than the previous higher multiplier . The lower the multiplier the less voltage it uses.

You need to put certain load to reach the multipliers between. The reason your only seeing 6x and 12x is when your idling and you load a program, it will automatically throttle to 12x (full speed of your CPU) so it loads faster then goes back down to 6x to idle and save power

Try scanning with an anti-virus, this usually puts medium load on the CPU which constantly uses the 8x multiplier or so

btw CPU-z was really made for desktops, thats why its displaying the wrong voltages and sometimes detects the CPU incorrectly

 

Flipfire: I am sure when I look up awesome on wiki, you will be right there. Thank you for your help and you answered all my nagging questions. I am on vista, but not using its power management.

 

hi there,
btw great guide flipfire!
I have a p-6831fx running off a T9300. I 've done a good amount of testing only on the 12x (highest normal multi) and I can only lower it from 1.1375v to 1.0125ish beyond that is BSOD (got at least 5 times). My question is when Orthos stops after 5 hours saying there's some sort of error, does that mean the voltage is too low on 1.0125v?
EDIT: here's the exact quote,
"FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 5 hours, 22 minutes 2 seconds - 1 errors, 0 warnings.
Execution halted."
i mean its not even a .150 decrease which according to this guide is where most ppl are at (plus i have a penryn which should go even lower than most). Please advise. Thanks!

 

Lowest voltage vary chip to chip, but lately ive also been seeing a relation between different notebook manufacturers

Does the stress.txt file say anything? The error might have been caused by something else. Just to be on the safe side, set it to 1.025v and try running the stress test again.

You were probably expecting your Penryn to go lower, but im afraid thats the lowest it will go. I didnt get much either compared to others, lucky bastards!

Did undervolting make a big difference between your temps? Or was you penryn running cool already?

 

I believe it is so Intel can release new penryn's with 25w TDPs that cannot be fully duplicated by undervolting. (Though at max power we'll be ~equal, the 25w chip idle will be more efficient than ours though.)

 

great stuff thanks

 

I couldn't find the stress.txt file anywhere i've done a few searches already. Any ideas?
Before UV'ing t9300 my max temps are ~68C.. @1.0125v i get ~58C and @1.0250v i get ~62C

 

It should be in the same folder where you unzipped ORTHOS.

Look... 1.0125v seems to be stable for 5hours, unless your gonna game at 100% full CPU speed for 5 straight hours then it probably wont give you any issues.