The "Undervolting" Guide

cool thanks

So I guess I should use v 1.53

 

Does RM have to be running for these to take effect? Also, if I remove the program will it go back to default voltages/clocks?

 

Yes and yes.

 

I wish to thank the OP for the guide he written 2 years ago... I've just started the undervolting proces on my Dell Studio XPS 1640 (with 2,4ghz c2d cpu) and already getting promising results.

One question remains, however, and it is quite difficult to go through 470 pages of posts to see if it has been dealt with before: my max multiplier is 9x. When i'm done determining the lowest voltage for the 9x multiplier, is it safe to use that same voltage for the 8x multiplier voltage and go down from there? The answer may be an obvious yes, but I just wanted to double-check =)

Thanks in advance, and thanks again for the guide!

 

Yes it's safe

 

Why did you enable all those power states in RMC and then tell W7 to use only the top one ?

 

Hi. I have an Intel T2370 and just followed your guide, as I ran tempCore while watching a 720p film and my CPU temps were at 90+ (I decided to check when I noticed the metal bit in the USB port burnt if I touched it too long) but I was watching it in bed and the bottom air vent was blocked I think.

With it unblocked, I got 86C max from HWMonitor after 10 minutes, and after changing 13.0x (max multiplier) from 1.25 to 1.15 I got 81C with ~45C idle

The thing I'm wondering about, is I have Toshiba Power Saver running with these settings:

Full power for when it's plugged in to the mains



Long Life for when it's running on battery (still able to watch XviD videos, etc. with no lag)



This is because the Long Life profile gives me 2h30m+ battery life while watching XviD videos.

Will RM interfere with that? Can the two run together? What changes do I need to make for the advantages of undervolting with the advantages of Toshiba PS?

Thanks for the guide.

 

They will run fine together, no problems.

I think your temps are high, maybe you should clean out the dust and change the thermal paste?

 

For your information, I managed to get the highest state (9x multiplier) from 1.1125V to 0.9625V without any problems. This is with a P8600 (2,4ghz) on a Dell SXPS 1640.

Also enabled "Auto-adjust intermediate states VIDs" as a quick solution for the other multipliers (8x, 7x and 6x). After that I did 45 minute cpu stress tests for all multipliers without getting a BSOD. Basically, this is a vote for the auto-adjust option as a quick alternative to manually determining the voltages for the non-maximum multipliers.

Thanks again for the guide!

Temp. diff on 100% load seems to be about 5-6 degrees C (65-67 down from 72-ish)... pretty good (especially for a free, easy and riskless 'mod')!

Gonna replace the thermal paste soon on both the cpu and gpu as well to see how that further reduces the temps...

 

Thanks for the response.

I'm just wondering, (how) will they conflict? RMClock manages power schemes, but so does the Toshiba app?

Should I leave my Toshiba app running with the current settings, and still do the RM Clock thing, or disable it all together?

Can I do what I'm doing with the Toshiba app, with RMClock (I don't really understand what RMClock actually does)?

Well, I would but it's a laptop and I don't really feel like taking it all apart

Thanks again.

Edit: I think I've got it now

I have these settings on RMClock, with TPS disabled



Is that right? I've only changed the 13.0x multiplier so far (from 1.25), is it worth working on the others?

Also, what's the difference between "Current" and "Startup" in profile selection?

Thanks again

Edit: Sorry, another question

Is it right that my 13.0x multiplier can go from 1.25 to 1.05 and it seems to be stable so far, so may be alright at 1v

Is that actually enough?

Also, it says go down .1 at a time, is this the same for all multipliers below 13.0x?

Edit Edit:

I undervolted 13.0x multiplier to 1.000v and ran ORTHOS for an hour, which reported no errors.

However, I was just looking on Wikipedia, and noticed that my CPU model voltage is 1.075 - 1.175V

Does this matter? I was going to try putting it down to 0.95 (the lowest you can set it to, and the one 6.0x runs at)

Is that a stupid idea?

P.S. Sorry for the loooooong post, but I just have loads of questions

 

I had toshiba power saver running with RMclock, never had a problem.

Toshiba power saver just changes the screen brightness and that sort of stuff.

Rmclock just keeps the CPU voltage low. They do different things so they can work together fine.

Current = Profile you want to be set right now.
Startup = What profile you want to be pre-set when you start windows.

To speed up your undervolt:
Click on "auto adjust intermediate state VIDs".
Then set your voltage for 13x.
Then uncheck "auto adjust intermediate state VIDs" and undervolt the rest of the VIDs.

Don't worry about getting the BSOD, just make sure you save your work.

I leave ORTHOS running and ever 10-20 secs I lower the VID.

 

Thanks for the reply

Don't Power Saver and RMClock both make changes to CPU throttling though? (In the XP Power Settings, I changed the Profile from Power Saver to RMClock, and it disabled Power Saver...should I leave Power Saver enabled, so it uses the PS settings for everything, and just RMClock for the CPU voltage?

Anyway, I just set everything to 0.95v and have been running an ORTHOS Blend Test for 25 minutes now with 0 errors.

Is there anything else I should configure?

Thanks again.

Edit: It's being going for over an hour now with no errors. I'll just leave it now and see how it goes (I don't mind reinstalling XP or whatever, I usually do once a month anyway), unless it's possible I could damage the laptop's hardware?

One thing I have noticed, is that it seems like the fan isn't running (it's REALLY quiet compared to usual )

I hope this is normal though, and I haven't just broken it seems to be sucking a lot of cold air in from underneath though, so I guess it's just not having to work as hard?

Anyway, thanks for the guide and the help

 

I had the power profile set to toshiba power saver.
So that is what I would suggest.

If I were you I would run intelburntest just once (only takes 10mins) to check that the CPU is fully stable.

 

I ran IntelBurnTest under Maximum and it was fine, but then I ran Xtreme Stress Mode and it crashed after about 20 seconds.

I don't really do much CPU intensive stuff on my laptop anyway though, the most is probably watching HD movies

 

Maybe give it a little more voltage, you dont want to BSOD when you didn't save your work

 

Hi

I used to have this installed under Windows Vista (32Bit), and had no problems, I have now installed Windows 7 (x64) and I am having problems with it.

I have installed the x64 drivers from the first post.

The problem I am having is that the program will not remember my settings when I quit and restart, or restart my computer?

Is there any remedy to this situation?

Thanks

Stewart

 

Hey, first of, thanks for the guide!

Second:
I'm stressing the CPU using Orthos. But after 2 (TWO!) minutes I've already reached a CPU temp of 90C and it's still rising.
I stopped the test by fear of damaging my system. This is not normal, I'm aware of that (hence my interest in undervolting).
As far as I understand I don't need to get that exact temperatue down so I'll skip this part for now... I don't want to take any risks :s

EDIT: I undervolted it by .100V and ran the test. After 5 minutes I'm still at 91C.

 

Clean out the dust, change the thermal paste, maybe use a cooler.

 

The fan and PC are dust free... Never looked into thermal paste but I may just have to... thanks!

 

hi!

im experiencing a bit difficulty undervolting my core i-5 laptop. As you may see in the SS, the RMClock doesnt recognize the cpu at all.
The advanced setting is also somewhat missing.
In addition, the HWMonitor also does not show my GPU Temp.
Can anybody help me please?

[/spoiler]

 

I think there is no solution for You. RMClock was written before any i5 or i7 hit the market and I guess it will not be updated. Try to find some other software.

 

You're not done yet. Don't report high temps unless you're reached your LOWEST stable voltage. Then, and only then, are additional measures needed.

 

I didn't go further in my tests. Undervolting wouldn't have gotten temps to safe levels I feel. I mean 2 minutes and it's at 93C? I wasn't going to take the risk.

I ran tests with the cooling pad on and it all works fine and I can play games without it shutting down. So so far, so good. Better spend 40 euros and be safe than having to replace the whole laptop because of a meltdown... :/

 

Return your "crystal ball" while it's still under warranty. It takes no more than a minute at each step down to find your BSOD point. Then, raise it up 0.0250 volts and you're pretty much done. BTW, CPU chips have a built in thermostat.. When they reach about 100C, they do the smart thing and turn off until cool again. Laptop coolers are a band-aid approach to flawed internal design.

 

try the CPUgenie trial version... maybe it may work... but its 30 day trial so u need to buy it once trial is over and if it works for u.

 

Well, the first of the two replacement laptops arrived. A practical clone of the first 1720 (stats in updated signature). Almost entirely the same with the exception of a slightly larger HDD and W7ULTx64 instead of the original's VHP32.

This one would't go quite a low as the .950v of the first one. Best I could get was 1.000v for long-term stable Orthos. .950v bluescreen'd almost immediately. .975 caused Orthos to stop working. 1.000v has been running for hours with no issues. Looks like this P8600 is not going to do quite as well as the last one.

Still made nearly a 10C difference in peaks temps down from 'stock' voltage on this P8600.

Looking forward to seeing how the P9700 undervolts when my E6400 arrives in a week or so.

 

Undervolting is supposed to lower your temperatures, not increase them.

And you don't have to worry about a 'meltdown' because cpu's have a built-in fail-safe to prevent this kind of a scenario.

If your temps reached over 90 degrees by just stress testing the cpu, then it's probably undergoing the same thing when put through it's paces in a game or a cpu intense software.

Oh, and despite the fact the cpu/fan might be dust free, have you checked the area between the cpu fan and the air vent?
This would require of you to unscrew the cpu fan and just move it to the side to gain access to the air vent.

The fan itself might be clean, but the vent area might not, and that could easily spike up your temps to the levels you're experiencing.

Undervolting cannot possibly result in damage to a cpu or temperature increase (unless you were increasing the voltages and not reducing them).

 

Just wanted to update on the situation I described a few pages back. My issue, using RMClock under Windows 7 64bit, was that my VIDs seemed to be bouncing back and forth between the default 1.2125 and the settings I made in RMClock.

I tried changing power plans, letting RMClock handle my power management, and just about every single thing I could think of or find on the innerwebz.

Finally fixed it, it seems: Apparently, the 'Minimum processor state' setting in whichever power plan you are using has to be set at 100% for some reason in order for RMClock to have full control over your VIDs. I confirmed that this was working as desired by comparing what RMClock was reporting to what was being reported by CPU-z before reporting.

Also, for anyone that is having the dreaded C2D whine issue (which, as I discovered after upgrading, can be quite annoying in a quiet room), I point you to this thread, whose solution seems to do the trick: http://forum.tabletpcreview.com/showthread.php?t=3516 YMMV, but it worked instantly for me. No more suicidal thoughts!!! lol

Anyway, just trying to pass on some possibly useful information for anyone else that has this issue.

 

No no let me explain:

Before I heard about undervolting, I couldn't play any games because the PC just shut down after about 20 minutes.
Then I started looking into temps and I got about 55 to 60C while IDLE.
I did a stress test using the software advised on this thread and under stress after only two minutes it had reached 93. I didn't push it because I didn't feel like it.
Then I tried undervolting it, only by .100 first, and it was at about 93 as well, after 10 minutes, and still rising. Which is an improvement, but it's not the solution I wanted: if it reaches 93C and doesn't shut down, then it'll be even higher when I play a game. Maybe it'll be just low enough to allow me to play, but I'd rather not risk it.
With the cooling mat everything is fine and the CPU/GPU are at a much cooler temperature. I know it's safe so I may experiment on it again, but right now I just found another solution to my problem.

 

I guess we're not getting through to you. You're not done undervolting until your voltages are all the way down to the minium level you can achieve on your make and model of laptop. The lower the voltage, the lower your temps. Why are you quitting before the job is done ?

I posted these choices a few weeks ago. Because of W7's stronger than ever power controls, you can do one of two things:

1. Use RMC for voltage control only and turn off plan integration. In that scenario, the min/max voltages in W7 must be set to 100% to give RMC control over the whole range. Then, do all the other stuff with W7.

2. You can choose plan integration which is not as good because RMC's copy of the W7 plan is incomplete and missing many components of the original.

I finally settled on #1 because both of them having simultaneous control over the same settings resulted in chaos.

 

Because I found another solution I have no need to undervolt now that my temperatures are back to a reasonably normal level with the cooling pad, simply enough.
But I guess you mean that I could have made a lot more difference by going all the way to my lowest level? I got it down by .100V and it was still way, way too high, which is why I didn't really bother with it once I figured out the cooling pad works.

 

0.1 volts isn't even halfway done for most of the hundreds of laptops I've seen here. If you had finished the process you wouldn't have needed to buy a cooling pad.

 

Must have missed that update. Maybe we can get flipfire to add that to the original post, since 7 is mainstream now?

 

Attention W7 users with Core 2 Duo CPUs. "Unclewebb" aka Kevin has posted the latest Throttlestop (ver. 1.94) at http://www.sendspace.com/file/o878u8

Set single CPU voltage and FID with this small utility and let W7 have its way with power planning.

Won't work for speedswitching "slackers"

 

followed the guide above and downloaded the rmclock, but when i run it there isnt a "Advanced CPU Settings" tab? and my cpu is "unknown" under the cpu info tab. help...

 

What is your CPU and operating system. RMC doesn't recognize names of newer CPU's but still works OK.

 

i'm having this problem too. the guide in the first post is out of date. i have an amd turion x2 processor, 2.2 ghz

 

For newer AMD stuff, try K10Stat. http://k10stat.dejadomain.com/k10stat.zip

 

Is this why on my 'new' W764 install I've noticed that sometimes the voltage is going higher than I've set it in RMClock? Been driving me a bit batty since I set it up this last weekend on the replacement Vostro.

 

Me too. I still get IDA spikes which disappear when the CPU is busy and reappear after a while at idle. To reduce them I did this.

1. Use RMC for FID and VID only. Uncheck just about every option box you see.

2. Disengage power integration.

3. Lower RMC priority to normal from Real Time.

3. In RMC registry tweaks, change all 1 sec. sample times to 10 seconds.

OR, use Throttlestop plus W7 power planning. RMC is dead. It doesn't behave with W7 and the new i3, i5, i7 processors apparently don't undervolt at all.

 

on my HP DV5 i only find 4 multipliers index 0 super LFM 6x ----index 1 normal 6x -----index 2 normal 7x ----- index 3 normal 8x , is this normal?

 

New E6400 came in today. P9700 (@2.66 down from 2.8 due to the way RMC handles half-multis) running stable so far at 1.0000v under orthos. Pretty happy so far.

CPU bouncing back and forth between 50c and 51c (down from mid-60s at stock voltage)
ACPI-"THM_" (chipset? or is that the thermal junction?) bouncing between 57c and 59c (down from ~70c)

Idle temps for the CPU are in the high 20s, which is quite nice overall.

I'll let it run overnight at is, and if stable, drop the voltage a bit in the morning and keep testing.

I'm going to be very surprised in this P9700 ends up being able to run at a lower voltage than the P8600 in the 1720 Running neck and neck at this point!

Anyone pull the HSF/heatpipe and do a AS5 treatment on it yet? Worked great on my M1330.

 

is this normal , i have only the max of 8x and on the 8x i can run on the min voltage without getting blue-screen of death ., what am i doing wrong all people here reaches like 11x or something , and i think no one did run the max multiplier on the lowest voltage without a blue screen of death , please help me , here is some screen shots

 

I was also able to set mine at the lowest setting and it remains stable. My stock peak was just over 80 before the fan started cooling, but then would hover at 75. Now my peak is 57. HUGE difference!

 

i think we are doing something wrong man , i did all multipliers with the lowest voltage and its working , i didn't even see a single BSOD , i am 90% sure there is something wrong , but i got a huge difference in CPU temps around 20c or 25c

 

For those who have Core 2 CPU's with half multipliers , to get full speed try the below fix in the thread link i'm attaching.. the pictures might not work but in the end even if u get a 16X multiplier or what , that will be ur max multiplier (for my T9400 , 16X multiplier , its 9.5X)...

http://forum.notebookreview.com/showthread.php?t=411168

hope it works.

 

Before my M1330 was stolen, it was running at max multi at the lowest available voltage (0.0925v if I remember correctly for the T6400 that it was equipped with).

There are MANY, MANY variables that can affect how low you can go, not excluding the core and stepping of the chip itself, its position on the wafer in the fabbing process, the chipset, other associated HW depending on the CPU, the BIOS itself, and the OS installed, among others. Even on a system configured identically to another, you can have a wide swing of underclocking results.

For example, my wife's last 1720 ran stable at 4 steps lower voltage than the replacement 1720 (with the EXACT same CPU, HW config, OS, etc., as the prior one). Sometimes you just get lucky with the right combination.

The P9700 in my E6400 is running at the same 1.000v undervolted as the P8600 in my wife's Vostro 1720 and I'm not done testing yet. It ran all night under Orthos at that voltage and today I'll be trying a few steps lower to see how low I can go. Maybe I'll be lucky and get down to min voltage. Doubt it, but results already are pleasing.

 

i don't know why , but i feel that i did something wrong , still my longest test is 15 min , but i am using the computer and its doing great , i ll try games in a few days to check

 

I ran the test for over an hour. I've also run the test while downloading, watching avi's, and browsing the web. Found a copy of Warcraft 3, and tried playing that and no bsod.

 

@brentbizzle so we can say we are toooooo lucky , i won around 25c less , i think i would get around 15 min more from battery