The "Undervolting" Guide

The T5750 seems to undervolt well for everyone. I only use a two-speed "transmission" on my A6120, 6X and 12X both at 0.950 volts. I found the load threshold box to be really important for me. I set it at 15% load so I get 2GHz power on demand, and back to 1GHz "loafing" when my application is finished. "6 on the floor" is for expensive sports cars. Just to have some fun, I ran Orthos on top of Winamp playing internet smooth jazz in the background! ha ha. Good music, no BSOD. What else could one want?

Permanent undervolting can only be done in the factory BIOS. Good luck there. I've looked for years just for a program that can access notebook fan controls, other than for Dell. I don't know why notebook mgfrs. have to cripple their BIOS's the way they do. Twice the price, half the fun. Geez!

 

Like I said, mine could not get below 1.025V at 12x. Is it because the rest of the comp is different or is there something I can do?

 

1. Why can't i lower my voltage passed 1,0625, its not letting me go lower than this,

Also under Advanced CPU settings

2. Do i need to click Enable Thermal Monitor 1 and Enable Thermal Monitor 2

and do i unclick Enable Enh. Deeper Sleep (C4E) and Enable Hard C4E

I noticed he did it to his in the picture but under his instructions it doesn't mention it at all. I was thinking those options are for those that have laptops that won't recognize their CPUs

3 I have a T8100 2.1ghz 800 fsb processor. What are your guy's voltage multipliers. I know they are different but i want to get an idea.

4. Do i run stability tests and lower the voltage of the IDA as well?

 

anyone? i've installed and uninstalled multiple times and reset multiple times..still won't show up.

 

@sciencefair, I'm not sure why. I thought maybe you had an older version but it's the same as mine. Maybe download it from a different source.

 

This is probably useless advise but close RMC run the regwipeout file from the RMClock folder (maybe reboot) and open the program again. Who knows?

 

Didn't work. I downloaded it off the creator's site, i'll try somewhere else.

anyone want to send me their .exe?

 

Could it be that RMC is not recognizing your CPU so not giving the page? What do your other tabs say?

 

good thread

edit: my spelling sucks

 

Is it possible to undervolt a T9400 with RMCLOCK?

 

Not so fast here. Watts of power used (all given off as heat) depends only on the voltage (pressure) and the volume flow of electrons (amperes) at that pressure. No different than a water pipe. At a fixed voltage, a CPU consumes the same amount of power (watts) regardless of GHz speed. If you can get your chip to run at 50 GHz at 1.000 volts, it consumes no more power than at 1GHz at 1.000 volts. There's the free lunch. IC's are not motors or light bulbs. The drill here is so simple: Run the highest speed you can, at the lowest voltage you can and still stay in the temp. range you want. If you can run top speed at the lowest voltage, then you can just stay there and s**tcan all those "clever" power schemes that confuse more than enlighten.

Who said RMClock is a resource hog?? News to me !

 

CPU Model says Unknown CPU but from what I understand the guide says if this happens you have to change something in the Advanced tab to get it to see it.

My laptop is a HP dv5z running on a turion zm-80 with vista x64 if it helps. And yes I did get the 64 bit driver lower in the page.

 

Will lowering the other multipliers decrease max. temperature even further ?

 

While the above is true, it is also true that a BSOD can, and sometimes does, result in corrupted files - critical, need to boot, system files.

Us Joe-blows with a few hours to spare and recovery disks can live with this small risk. Messing with voltages on a more critical system is better left alone.

http://forum.notebookreview.com/showthread.php?t=293485

 

Interesting quote. I implemented above suggestion; and go a bit further. See graph attached.

I do a short test. I use the same voltage for two different CPU frequencies and place it under 100% workload to extract heat emited from the CPU. At 997MHz, heat recorded are at [60, 57]'C, while at 1663MHz, heat recorded are at [68, 65]'C. If heat == wattage, I suppose CPU time would matter too.

 

Errr, every time the clock ticks, a signal pulses across your IC. Switches open and close, energy is converted to heat. Higher clock speeds = more heat. As electricity is being converted to heat, and more heat is generated, more energy is being consumed.

Where is the flaw in this?

 

Sciencefair: Show us your "CPU Info page" next time. You have the latest RMClock, 2.35, but your CPU may not support the advanced features of the program so the tab won't appear.

Spiral: RMClock doesn't take ANY resources. It simply replaces the stock power settings of the OS in memory with your own settings. Something wrong. I personally wish it would permanently overwrite my bios and acpi system files but it won't. Look at your startup files with Taskmanager or the like. RMC runs, loads your pref's, and then quits until next boot or until YOU change something else.

 

Thank you to the couple of people that caught me over-simplifying basic electronics. I kind of glossed over an important point with IC's. They're not simple resistors like a toaster, nor are they inductive motors as found in a hair dryer. They're electronic switches. They require a minimum voltage to turn on and off, but not much current to do so. So, at idle (no load), power consumption (heat, watts) doesn't change much with frequency. More simply, flipping a switch on and off twice as fast doesn't contribute much to the overall heat production of a CPU. But the killer here is the volume of current being switched around a tiny siicon wafer. Remember, a really small switch can control a lot of amperes. At high loads, CPU'S are moving around a lot more amps. There's the bulk of the generated heat. Undervolting reduces the pressure (voltage) of this electron flow and so reduces wattage directly. I hope I never implied that undervolting leads to superconducting. Yikes !
The 2nd Law of Thermodynamics still holds: No free lunch. Thanks for catching my undervolted brain idling.

 

Not trying to jump down your throat, Scott. Just trying to keep the info clear.

RMClock is always running, you can see it in task manager. It also installs a non plug and play driver called RTCore32, which can be seen in device manager if 'view hidden devices' is selected. The driver is what requires admin privileges in order to use RMClock. Don't know if you ever tried to run it from a limited account, but it just wont work.

When RMClock is shut down it uninstalls the RTCore32 driver (shown grayed out in device manager when uninstalled) and the computer uses the default CPU driver alone.

 

Sorry for the inactivity in here, been busy.

Im afraid the processors in the new Puma plaform isnt supported by RMclock. The RMclock development team has died out.

Yes but it will downclock to 2.4ghz because RMclock cannot recognise half multipliers. T9400 has a 9.5x multiplier, it will read it as a 9x multiplier

 

1. Why can't i lower my voltage passed 1,0625, its not letting me go lower than this,

Also under Advanced CPU settings

2. Do i need to click Enable Thermal Monitor 1 and Enable Thermal Monitor 2

and do i unclick Enable Enh. Deeper Sleep (C4E) and Enable Hard C4E

I noticed he did it to his in the picture but under his instructions it doesn't mention it at all. I was thinking those options are for those that have laptops that won't recognize their CPUs

 

Looks like you're still using the CPU in desktop mode. I believe it's under the advanced tab where there should be a choice between mobile and desktop, choose mobile and apply. (Only read the last few posts from previous page so I don't know if you posted something else before)

It doesn't matter, you could enable thermal monitors but it won't make a difference in how your CPU performs. You may as well leave C4E checked since it's a really deep sleep setting for saving even more power, unless you don't want that that is.

 

sorry if this is a bit off topic, or has already been posted, but has anyone tried undervolting a P9500 yet? If so how significant was the volatage/temp drop?

 

I've dropped a P8600 from 1.1375V to 1.05V and successfully run an overnight stress test. it died, however, at 1.025V. The lowest voltage for my P8600 is 0.925V and that is working fine for the 6x multiplier (1.6GHz). RMClock can't handle the half multipliers (one reason why I chose the P8600) so the top speed you will get out of the P9500 is 2.4GHz.

At 1.05V the maximum temperatures were down about 5C and power consumption reduced by 3 to 4W. One core of my P8600 is significantly cooler than the other:



It's a shame the second core doesn't match the first - I'd have a nice cool CPU.

John

 

So, I followed the undervolting guide precisely and the temperatures that RMclock and HWmonitor are giving me just seem a little unbelievable. My max temps are 38C while my mins are 32C. I keep my apartment cool...about 72F.

From looking at other people's max temps, it just seems to me that the temperature that its giving me couldn't be correct. I've been running Orthos for an hour and a half now and the temps refuse to go any higher. Do you guys think both the programs could be wrong? or does my computer just run extremely cool?

My computer is a Gateway p-7811fx, chipset is a P-8400 2.26GHz core dou

I have a notebook cooler but I turned it off to put more stress on the cpu and get those temps to go up. Also, the task manager confirms that my cpu is at 100% load.

 

I just turned off Orthos and my temp went down to 27C in about 15 seconds, can it really cool down that fast?

 

A few gateway users have had problems with temperature programs. It seems to detect it 30c cooler than its supposed to be. I think it might be a BIOS issue giving out wrong signals.

Its a gateway specific issue. Diablo85 and Lajkin has also experienced the same problem.

 

I have undervolted my Dell M1330 with a Penryn T8300 (2.4GHz), and it all works far cooler now (all on 0.95 V, except last two (12x and 13x) which are at 1.00v and stable).

However, when using AutoGK to convert DVDs to DivX files, I noticed that RMClock sometimes drops the processor to 600MHz, even when doing the video encoding (first pass). Does this mean that RMClock is not realising that there is a high CPU load, or don't I need to worry?

It seems like the encoding is taking maybe 5-10 longer than it used to (for similar files), so not sure. Thanks.

 

Geez........
Look at all the trouble we have to go through because manufacturers of $1,000+ notebooks try to save a couple of bucks on cheap fans. A great guide, 153 pages of questions, thousands of man-hours tinkering and still counting........
Now that my laptop can't fry itself to death, I'm going back to computing

 

so for my config, i am using .95 V @ 6x and 1 V @ 12x. I never went lower, but i'm curious to see if anyone else with the same processor/ram combo is running with lower multipliers and stable ?

so far i've not encountered any BSODs. just checking in before i lower the voltage even further.

 

With my TL-58, I have the following voltages:

4.0x ..... 0.7250V
9.5x ..... 1.0000V

My temperatures (with HP's new BIOS) are 48C idle, 63C max.

Great guide, Flipfire.

 

Did you enable throttling? Turn it off. In the CPU business, throttle means slow down, not accelerate, as with cars.

 

I tried this before on a Fujistu P7230 but i decided to give it another go as maybe i might have done it wrong before.

Will this even work properly on a 1.2 Core solo processor?
I am asking this as the RM Clock will not even allow me to change some of the settings on this page.

Edited to add:
Well i tried it again and lowered my voltages to as low as they would go and it does not appear to have made any difference.

 

Can you see your voltages and multipliers?

Some of those settings are automatically detected and set like on my Pentium M.

I suggest you tick "Apply these settings at startup" and hit Apply

 

Flipfire thanks.
These are my default voltages and multiplier settings.
I also tried unticking the last 2 highest multipliers after i lowered the voltages as well.

 

The U1500 in my Sony G11 has a voltage range of 0.9375 to 0.9625, so the bottom voltage is the same as yours and the top voltage is a little lower. I put everything to the lowest voltage and ran a dual core Prime 95 stress test to confirm it was OK. The G11 stays nice and quiet under light usage with the CPU running at 800MHz most of the time.

Personally, I would never call 0.9V at "ultra low voltage". The grandfather of these CPUs, the Pentium M, could run at 0.75 to 0.8V and my T7300 spends most of its life cruising at 0.9V.

John

 

As it worked on your computer ok John that also uses a Core Solo processor i don't know why it never worked on this P7230. I agree with you this processor is meant to be a ULV Core Solo processor but like you said more powerful processors even have lower voltages.

Maybe the fact that this laptop is 10.6" does not help either but laptops as small and more powerful than this even have better fan control than this laptop.

I doubt if i will able to get Fujistu to give me a refund on this laptop. It is due to be collected tomorrow and i have asked them to look at the fans again as it was only booked to have the lid replaced. Maybe more thermal paste on the CPU might help i don't know. Under light conditions now just browsing on Firefox with ad block plus installed and no script the fan is running very frequently.

 

I've not pulled it apart but I suspect the G11 may have a size bigger (and therefore quieter) fan. There are also design issues such as the temperature at which the fan operates (see the RMClock monitoring page) and how much heat gets transferred into the chassis. More thermal paste won't help. In fact it will make it worse since the layer of paste should be just thick enough to fill any gaps between the CPU and the heatsink. Any more acts as an insulator and will increase the fan activity because the CPU gets hot but the heat isn't being removed. Does the fan blow out hot air when it is running?

John

 

John the fan does blow out air of varying temps. I will forget about the thermal paste idea then. I don't know what else can be done now.

 

Wow CPU-z also cant detect Xeon voltages properly



0.048v!! amazing

 

I have a T5250 as well, and have the same settings, same result. I went from ~70C to 57C. Maybe it is the processor? (I stressed it for more than 4 hours on the lowest possible settings)

 

how did u get ur HDD temp to appear in HWMonitor ?

 

Welcome to the club. You're supposed to get those results with a Merom.

 

You must have a special HDD with S.M.A.R.T.

 

i ran orthos for 4 hours last night, no BSOD. i set my 12x @ .975v and 6x @ .95v (everything in between is between these 2 values).

is it safe to say i've got a stable system with these settings ? i didn't see much of a drop in temps, as my processor is still idle @ around 35-37C and still tops out @ around 55-57C.

i wonder how fast my machine would be if i zip-locked it in a freezer bag and left it in my fridge overnight then fired it up in the morning....

has anyone invented a refrigerated notebook cooler yet ?

 

make one yourself ^^. But being cooler wouldn't raise the speed...

 

i want to...but what i have in mind would require funding in the order of a typical gov't funded program...he he

but it would be thinner than a macbook air and require no external power

 

Hey guys, what's going on. I previously had my processor undervolted moderately, but I'm trying to push the limit (I need battery life for class, even though mine is fine now, why not try for more!!)

Here is my latest attempt - everything is 100% stable.









Check out all the voltages for all multipliers - I've undervolted them all. Do you guys think I could go much further? Would .9625v work with my T8300? Also, at FULL LOAD I'm running about 54-55C. Good/bad?

Thanks!! Any feedback/comments are welcome

EDIT!!

 

My HP8510p with T8100 can be undervolted to .9625 for all multipliers and be completely stable. Temp at idle is around 25-30 degC. Max temp when gaming is 40. And that's after modding ACPI to make the fan quieter.

Unfortunately I'm getting a BSOD once in a while - made me think undervolting was crashing it, but not so (wouldn't crash during stress tests but does after ~12 hours of use no matter what I'm doing). So the Penryns do seem very flexible with voltages.

[ed] Found the cause of my BSOD. I'd previously patched my ACPI to adjust fan speeds, which worked great until I decided to update the BIOS. From that point, BSODs. Reinstalling XP fixed everything, no probs since then. Lesson: restore default ACPI settings *before* patching the BIOS, then re-patch ACPI.

 

Looks fine You can keep on trying to lower your 11x multiplier till it BSOD's, then you will know your limit. I suggest raising the voltages by 2 steps higher when you have found your limit.

Make sure all uneccessary programs are closed and work saved when attempting to find the limit.

-Enable SuperLFM if you want more battery life. On idle, SLFM will cut the FSB clock to half.

Btw, you can get your core temps to show on your taskbar/system tray. Right click the RMclock wheen icon and look for 'Show CPU temps'