The "Undervolting" Guide

thanks andy. got started on the new voltages. .8750v was 45 min orthos stable, down to .8500v now and testing. temps holding at 43 degrees Celsius.

 

I like CrystalCPUID's menu better, but I still can't figure out how to set it up, any advice? I simply want to reduce the voltage at both x6 and x9 multipliers (I have Intel SpeedStep enabled). I don't want to change the multipliers or anything else, just decrease voltage. I don't want to add a third "middle" state. I have a p8600. I've attached the picture.

 

RMclock hasnt been working out well for the newer processors.

I suggest just waiting for the new NHC to be released. This will have better support for both Intel and AMD mobile cpu's with loads of more power features.

We have received word that the development programmers of RMclock have gone AWOL because they do not wanna work for free.

 

Functions > Intel Enhanced Speedstep Control > Select whatever multiplier and voltage, you want active....

 

Are intel chips hard coded to stop lowering the voltage once it hits a certain point? After i put UnlockVid in the rmclock registry i was able to open up more voltage options but nothing below .9000 (which is 4hr orthos stable) has any effect. no temp change, no stability change. nada. Am I doing something wrong?

 

Thanks for this great guide! I've successfully undervolted my T8100 down to a mere 22 degrees Celsius on idle, down from 33 degrees Celsius, and peak temps were formerly at 56, and now stand at 44.

My laptop is a 13.3" LG P300 with already low temps, so bringing this down further really shaved off some power consumption.

Voltages were at 1.1375 through indexes #1-#10, and I disabled the IDA index. They are all now at 0.9500V, with Index #0 (SuperLFM) at 6x and also at 0.9500V, running at 600MHz on idle from 2.1GHz on Peak.

 

Yes, Intel has locked the minimum voltages after Pentium M's to stop idle instability. Unlocking the voltages through the registry will have no effect whatsoever

 

What's NHC??

 

Notebook Hardware Control. It lets you control all sorts of hardware in your notebook. Its got some cool features, undervolting is one of them

The version thats currently out only supports XP and older CPU's. They are currently developing a newer version

http://www.pbus-167.com/

 

So, with a T9300 chipset running XP, I should wait to undervolt?

 

Well it will work with the T9300.

Except the T9300 has a 12.5x multiplier. RMclock doesnt support half multipliers. It will detect it as a 12x multiplier meaning it will downclock by 100mhz.

12 x 200mhz FSB = 2.4ghz

Instead of 12.5 x 200mhz = 2.5ghz

There is a way around this by enabling IDA

If you had a T8300 then RMclock will work perfectly

 

cheers flipfire

shaved 15c off idle and 15c off load
(without really pushing the voltages)

worked a treat

 

RMClock does not identify the P7350 only the 6 and 7 multipliers are shown with this.
Are there any other tools that we can use for undervolting the P7350?

 

As said above, RMClock will not recognise half-multipliers. The max multipler of the P7350 is 7.5x, and a 133MHz loss per core isn't gonna make a huge difference.

If you have XP or Vista 32, you can try out NHC, or CrystalCPUID, but I am not sure if they support half-multipliers.

 

my T9400 experience a throttled speed from 2.53 to 2.39 ghz when i undervolt it from 1.15v to 1.05v .. my maximum multiplier is 9.0x.. so i presume i did underclock my t9400? is there anyway to solve this auto throttle problem?

 

It is buggy, and the 0.133GHz decrease in frequency is not going to affect performance....

 

I'm know I'm really late to the party and should have done this months ago, but I'm currently using RMclock to get my CPU temperatures down (98 maximum was the absolute limit).

How long does it take normally for errors to start showing during Orthos stress tests? My CPU is identical to flipfire's and I'm now on 1V flat. My last 2 stress tests were for 3 cycles whereas they were initially on 2.

So far results have been impressive, max load CPU temps are now below 75 whereas early stress tests would force the CPU over 100 and shut my system down by force.

 

Some say a 24 hour orthos stress test means a system is going to be pretty much rock solid and never crash. I find 1 hour of orthos good enough for me, i never encounter crashes with what i use from this (games, music production).

 

Not long after I made my last post I had a BSOD midway through the 3rd stress test loop, I guess 1V flat is too much for my CPU to take.

I'm chancing my arm and keeping the 11x multiplier on 1.0375V (2 above the last stable 3 loop test result) and lowering all the multipliers by 2 choices from 11x until the minimum. Is this a really bad idea?

RMClock for me currently shows this:

SuperFLM 6.0x 0.85V
7.0x 0.925V
8.0x 0.9625V
9.0x 0.9875V
10.0x 1.0125V
11.0x 1.0375V

 

Which ever works for you

The old undervolting stability test with NHC was only 15mins lol

Its best you let RMclock automatically adjust those for you. Do the following:

1) Click the Default Button
2) Tick "Auto-Adjust Intermediate VIDs"
3) Choose 0.850v for your 6x multiplier
4) Choose 1.0375v for your 11x multiplier

RMclock will automatically fill in the voltages in between those multipliers.

 

I've done as you suggested and the calculated settings are actually more ambitious than what I've selected, I'm hoping an overnight test will prove stability. After this if it works all I have to do is make the program auto-start at the "performance on demand" setting, or is there something I've missed?

 

It should be fine. Try running the usual programs you use and if it doesnt crash then its fine.

An overnight test will only stress the 11x multiplier. You will need to individually test 7-10x for 1 hour each if you really want a full stability test.

 

That's going to take a while, individual testing of 7-10x multipliers is a good 4 hours (1 hour is roughly 4 cycles). I'll give this setting a couple of days and do a proper test cycle when I have time. Thanks for the advice.

 

Yes your intermediate voltages are most likely fine. Just use your notebook normally for a week or two, if you dont get a BSOD then its all good. The 7-10x multipliers dont really get used much. Its usually just Idle and max multiplier. If you start getting BSOD's thats when you start testing those multipliers.

On my other notebooks, i usually just make sure the max multiplier is fully stable and let RMclock adjust the intermediate multiplier voltages accordingly. and it worked out fine.. too lazy testing the other multipliers.

 

Sounds good, I'll save a bit on the electricity bill by passing on the overnight test. One last question, is there a specific error message to look out for on the BSOD that will hint at undervolting being the cause? The only BSOD I've had so far was the 0x8000000E or F when I was in the middle of testing 1V.

 

Is there any way to link RMClock profiles to the various Vista Power Plans. Ie: when I select "Balanced" RMClock will flip to the "Power on Demand" setting or if I select "Power Saver" it'll change to "Power saving"?

 

I have just got my clevo m860tu gaming laptop and im just a bit confused.
I ran the orthos Stress Test and monitered the temps using Hardware Monitor and after 14 mins my CPU's max read 31degrees in core 0 and 28 on core 1.
You said to expect results in the 70s or 80s.
I used both the Orthos blend test were it tested both CPU and RAM as shown in the thumbnail as well as just the CPU test.
Am i doing something wrong or can i continue with the Undervolt Guides next step using these temps as my CPU's maximum temps.
Just for the record my GPU (9800GTS) read 39 after the 14 min test and my HDD read at 58 which is really high i know but the M860TU is just like that for HDD temps as there are no vent holes there. Everyone has reported pretty high HD temps.

 

i have a question about superlfm option. so at 6x multiplier on normal, the voltage is already at its lowest at .9000v and the frequency is at 1.68ghz. with the superlfm setting on, it is also at .9000v 6x but at .84ghz. since i cannot lower the voltage anymore or the multipliers, is there any actual benefit to using superlfm option?

 

SLFM mode will also reduce the FSB to half.
so you CPU will be running at 0.42Ghz, and also other components will draw less power (southbridge, RAM(?)).

 

first of all thanks for the guide, I've been testing the voltages for some time, and seem good, I have a a dv5t p8400, and I've managed to reduce voltages to 1.050v in 8x from 1.200v, 7x 1.012, 6x 1.000v. Are these good numbers? or you would say I continue till I finally get some error or bsod in the 3+hs test?

 

Thank you for clearing that up, I always had that doubt, since it used the same voltage. But if the lower multipliers used the same voltage, I should enable only the highest, right (without slfm)?

 

Now I have noob question. When I lowered voltage to value X at max CPU speed\multiplier may I assume that it will work stable with this X voltage on lower speed\multiplier ?

 

Seems logical but why assume ... just try it.

 

Yes, it can only be more stable as you lower the clockspeeds with the same voltage

 

Yes, you can assume a stable voltage setting for a certain speed will work also at slower speeds too. You may want to lower the voltages for the slower speeds even more.

Cheers,

 

Thanks. I'm in process of doing that.

 

Theoretically yes. You will be keeping the processor in C3/C4 states for longer the higher your multiplier is. The end result is you should be using less power,
and also keeping your CPU cooler under normal load conditions.

 

my friend's laptop has merom 1.5ghz and the lowest i can go is .9500v on all 4 multipliers. is there a way to lower it even more or should i do what is suggested above and turn off all lower multipliers?

 

No way to lower it further. Intel has locked the minimum voltages.

I suggest keeping all the multipliers enabled, you only have 4 anyway.

 

hey flipfire, great work on this tutorial man. I tried to rep you but it said I had to spread it around, lol.

I've been MIA for a while, you've done good man.

Remember a while back when we were noobs to u/v and tackled it together? lol

I had no idea about the signed 64bit driver til I saw this thread just now, I've been using your method of Task Scheduler this whole time =P

 

oh hey dude.. long time no see, no need to rep

but yes, it has been a while.

Finally acquired signed x64 drivers..time to get rid of the old method and use RMclock normally

 

i've tried to undervolt my Dell XPS M1530, T7700. but when i reach certain Voltage, my Orthos stops responding. I rarely see any hardware error and no BSOD at all.

May i know what is wrong? or is it just normal?

 

If its crashing when it pass a certain voltage then it might be low. Try using a different program to stress the CPU like IntelTAT or Prime95. See if you get any BSODs or errors.

Never actually seen ORTHOS crash like that before.

-

On other news, ive revised the guide on how to lock your CPU to full speed

 

i just used the max performance the way u described flipfire, and i notice that the core temp stayed the same as before at 36 degree c. i hypothesized that it might get hotter... considering 6x was on .9 v and 9x is on .9625v. im very happy that it didnt go up, but y didnt it?

 

It normally rises by a few degrees higher. The difference between my 6x and 12x voltages is big (.850, 1.1v) . Difference on yours is quite small.

Theres not much load being put on it so you wont really notice much of a temp difference. Remember the CPU mostly creates heat when it is processing stuff under load.

Try putting abit of load on it, see if it drops back down to the same temps

 

First of all, great guide!

Second, I'm trying to apply this to my T8300 in a studio 15 but it seems none of the multipliers will go below 1.0000V.
It's not that I can't select a lower option, just that under "CPU info" the voltage remains stuck at 1.0000V. Even for the lowest multiplier (6x) where the default voltage should be 0.9000V or something, it always shows 1.0000V.

I read another user's post with the same problem here a while ago, but can't locate his post now, so maybe it has already been answered but there are just too many post for me to read them all.

I checked the mobile box and did all other stuff as described. Any help on what might cause this is greatly appreciated.

 

i've just tried using Intel TAT to stress my T7700. the temps run much higher than Orthos. However, Intel TAT indicates my processor is Pentium M. is it correct?

 

Great Tutorial!! Thank you very much!
Mine is too a T7500, could you tell me what are your values for your multipliers, should be identical for me. I know you probably already post them but it´s a 200 page thread.

Thanks

 

Hmm it could be locked via the BIOS or something.

Try running RMwipeout.reg in the rmclock folder to clear all the settings and start again. Make sure you remember the votlages.

How much higher are we talking about? What temps are you getting?

The optimal voltage will vary from chip to chip. Though most T7500 are able to reach 1.0v to 1.1v. You will need to find the unstable voltage and make sure you put a margin on it.

I suggest ticking "Auto Adjust Intermediate Multipliers" so all the other multipliers are set automatically.

 

I was the guy who asked this question. It's not been answered yet so far. What I think is that Dell or Intel locked the processor. No matter what, I'm still not able to go lower than 1.00v.

Here's the link to my older post: http://forum.notebookreview.com/showthread.php?p=3915385#post3915385

I tried it but to no avail. Maybe because RMClock doesn't support newer batches of T8300? Just a guess. Thanks for the awesome guide anyway.