The "Undervolting" Guide

Unfortunately I only get temp savings and maybe a couple dollars less on my electricity bill since my battery is dead and I get maybe 10 minutes from it. I'm on an old Dell laptop that's being replaced later this year as soon as I get home from being abroad. The laptop I'm planning to get has an i7, and from what I've read there's no way to undervolt it like this.

 

You obviously have a guilty conscience about a successful undervolting experience. Sending me your income tax refund check will ease your distress, I'm 90% sure.

 

I know how he feels though. The way the tutorial is written the steps that should be taken are a lot like overclocking, where you try a little bit, test for stability. Makes it sound like the computer could BSoD at any moment. But when I was doing it everything seemed fine with each step, so just to get it to BSoD I jumped to the lowest voltage, and nothing. Everything was fine.

I was also going to post to see if everything was done correctly, but as I checked all my settings it seemed fine. Under load the PC even felt cooler so I knew that the results had to be real. I just wish it were possible to do this with the i7 so I could undervolt my future laptop.

 

Yes, the guide is old and conservative. A BSOD does not hurt anything, it's just a signal from the CPU that the voltage is too low. Solution: Reboot and raise the voltage a notch.

Unlike overclocking, which CAN fry a CPU, undervolting cannot damage anything. Most experienced users find their undervolt settings on the fly during a single and short stress test.

 

It wasn't that I was worried about damage... I just figured I'd see at least 1 BSod. When nothing was happening that's why I just threw on the lowest voltage, 0.950, expecting a blue screen but nothing. And it's totally stable!

 

Many others have reported the same lucky CPU-BIOS-MOBO combination giving a low flat energy curve. It means equal power use at every FID, which means speed switching saves nothing anywhere.

 

i experienced the same thing , finally i feel that i am lucky with my laptop

no no , I am completely satisfied right now , knowing that this can happen to people , i just felt i am not the person that good ( lucky ) things happen to them

u ll save energy as u will be running on highest multiplier using the voltage of the lowest , and getting more battery life though u ll be using highest performance

 

Pardon my question if it's been asked before (then I would need to go back and re-read the pages I've read) but does this guide still apply for the new Core i5 and Core i7 mobile processors? I've read somewhere that they can tune down and clock up when needed, depending on whether a notebook is on battery or AC. Or is undervolting still more effective than Intel's build in dynamic clocking?

Also, how does this affect Turbo Boost of the Core i5/i7 processors? Besides the fact it'll probably be slightly more challenging to find the sweet spot for undervolting due to the varying clock speed when Turbo Boosting, any side effects if someone undervolts their processor and the processor decides to boosts its clock speed up when running games/heavy software?

 

I'm having troubles to run my cpu with SuperLFM enabled. Everytime I try to set it under "performance on demand ", as soon as I hit apply my laptop freezes/hangs.

Does anybody know why? I already searched high and low with no result. I'm running seven x64 BTW.

And, I finally figured out how to lower my CPU voltage below 1.000 V. Enabling DFFS does the trick, like another member with the same problem already confirmed. I don't know how/why I missed that. I suppose my upgrade to seven will have nothing to do with it, but still, strange.

Too bad it just lowers the minimum to 0.9000 V. I have a feeling more could be shaved off since multi's 6-10 all run stable on said voltage.

 

I'm using a Core 2 Duo L9400 but the lowest voltage that shows up in RMclock is 1.0375. Cpugenie has 0.925 and passed on all the multipliers. I know I can get it even lower. Does anyone know how I can get the voltage to below 0.925 in Cpugenie and RMclock? Thanks.

 

i am having the same problem i reached the min voltage and i wanna try lower voltages

 

I don't know all the technical reasons why, but from what I've read this method to undervolt does not work with the new i_ processors. The settings made in the applications won't actually apply to the chip. Sucks cause the reason I started undervolting my current laptop was to practice before I got my i7 laptop.

Woot! I'm running the lowest which is only set at .950. When I get home tonight I'm gonna look for that setting and hopefully take my cpu lower!

 

Could you please elaborate on what you mean by this? Like what do I need to do? Thanks!

 

For me the DFFS option is grayed out. Is there some setting that needs to be enabled or is my processor not compatible?

 

Sup all!

I’m running Vaio FW 190E / P8400 Core 2 Duo Penryn @ 2.26GHz / 4GB ram / Win 7 Ultimate

My first heavy CPU load temps where around 62- 65 Celsius. I know that it’s somewhat cool temperature (especially because my old Amilo was scorching 80oC after one hour of playing 3D games)… And Vaio is abnormally quiet anyway… but since I like to tinker whit hardware, I’ve decided to cool it a little more (but wanted to keep it’s mobility, so thermal pads where out of the question). Since I didn’t know much about undervolting, I’ve decided to do it old fashioned way and replaced the stock “thermal bubble gum” that was on my heat sinks, whit Noctua HT-N1 paste, polished the contact surfaces of the sinks and had to fashion out couple of copper plates because there was a significant gap left between the sink, GPU memory modules and the chipset…. and and got only like 3oC decrease… that sucked, especially that I’ve broke the wireless ON/OFF switch during assembly :/

So I decided to study that famous undervolting a bit more, and after only an hour and a half of research and implementation of newly acquired knowledge : ), I’ve managed to decrease max load temperature form 60oC down to 44-46oC… and my idling temp from 40ish to nice and cool 28.
I’ve turned down the core VID from default 1.500 to 0.900, on my max multiplier (8.0X), did a Orthos stress test with the mode set on blend for an hour, and didn’t got any errors or warnings.
Default value of my superLFM is 0.875V, so I presume that those 0.9V are pretty low enough (can go lower thou).

Software used: RMclock v2.35; CPU-Z v1.53; HWMonitor v1.15 and Speed Fan v4.4 for temperature control.

Enough whit the intro … Point is, stay away from hardware modding kids, it’s not worth it : )

Now the question!
1. I’m hearing a crackling sound from time to time… I ‘m not sure that I’ve heard them before, but maybe I’m just tripping : ) All my drivers are up to date.
Any ideas?

Great forum, was VERY helpful! XD

 

What do you expect to do with your notebook while running on SLFM ?

Standby uses less than 3 watts total, gives a full 24 hours on a single charge.

 

Just wanted to add a quick reply:

I have an HP DV6000t with the infamous Nvidia black-screen-of-death problem. After reading all the stories and the sights with literally THOUSANDS of dead laptops, I figured that HP/Nvidia have no plans to unbrick my notebook with a free repair for a manufacturing defect. So, I went the reflow-oven route. I read about commercial reflow ovens and profiles, and tried to match those as best as I could. I pulled the motherboard, insulated where necessary, popped the MB in the oven. I did: 4 mins to/at 350F, 4 mins to/at 385F, 4 mins to/at 410F, and 2 min at 420F, 30 minute cool down. Commercial reflow ovens have better temp control, and can run this profile in about 5 minutes.

At the same time, I had wanted to upgrade my T2050 to a T7200 since I bought this laptop. T7200's were $300 three years ago! Now , though, they are $60 a pop brand new on ebay (if you have the patience to wait for one). I bought one several months ago in anticipation of a Windows 7 64bit upgrade.

So while I had the MB out, I installed the T7200 as well. I used Arctic silver on all of the processors, put it all together, prayed and hit the power switch... up it comes.

Now, the problem with the Nvidia GO 7400 in the first place was heat. The T7200 could only make things worse. I started up the laptop, looked and CPUID, started orthos and PANICKED! Temps were 93C and climbing! I shut down orthos immediately and started researching. I came across rmclock, and this thread. After a bit of research and experimentation, I now have my DV6000t/7200 at 12x and 1.0000 volt, interpolated down to 6x and .95v.

The 12x multiplier actually went all the way to .95v before BSOD, so I could go even lower BUT the biggest temperature drops came when getting down to 1v.... steady state 82C when running Orthos. I know 82C sounds high to most everyone, but it is not unusual for DV6000 and T7200, and it is WAY better than 93C and rising!

 

Very nice. 82C doesn't sound bad at all for an older part, and down to 1V is good as well. With my T8300 (45nm) I can't go any lower than 0.975V without BSOD'ing. Ah well, still down from the original 1.250V.

 

Hi,

really great tutorial, I stumbled on it when looking for ways to lower the GPU temperature in Dell XPS M1330.

I started following the instructions and I noticed that RMclock's CPU info shows Current Multiplier (FID) as 6.0x - which is my minimal FID. The Profiles page lists the following FIDs: 6.0x, 7.0x, 8.0x, 9.0x. All of them are checked. The CPU info's current FID stays at 6.0x even when I run Orthos tests.

From what I understand the FID is supposed to be dynamically changing based on the needs, so during a test, it should jump to the highest one.

Can anyone tell what I may be doing wrong?

 

@radoslav maybe ur running power saver profile that forces your notebook to run on the min multiplier , just maybe LD

 

@beige

Haven't thought about that Still, after checking, I'm running the Dell Recommended power plan, with "processor state" set as 5% min and 100% max. I guess, these are the relevant settings

Still, thanks for the suggestion!

 

From what I understand you need to change the processor state to 100% min in the windows power plan (at least on windows 7 you do).

 

From a configuration (rather than OS) standpoint:

There is the Profiles page, and that is where you set up your voltages on each index. Check them all, set input voltage(s), click "Apply"

One the Profiles page, you also set your Current and Startup management profile. Set "Performance on demand" for Current for both AC Power and Battery. Click "Apply"

Now, go to the "Performance on demand" profile. make sure the "Use P state transitions" is checked for AC and Battery. Click "Apply". Then, click all of the indexes in each box. Click "Apply".


On the OS (Windows XP), I went in to Control Panel, Power Options, and set "RMClock power management"

At that point, I could watch the on demand voltage/frequency changes

 

Does anyone know if this program will work with an i3 processor? If not, what should I use? I don't want to mess up my machine.

 

Apparently, it doesn't. Laptop components are now lighter and cooler and better designed so Intel doesn't need to incorporate powering down to protect their CPU's from overheating and cutting out anymore. It was fun while it lasted.

I don't recommend power plan integration with W7. Let RMC do voltages and W7 do the speed switching and all the rest. For this to work, you must set the processor states (in percent) that you will be going to want to use ie. 100% to use the top FID.

 

Can anyone with a studio 15(35) or any other Dell laptop verify whether they can run their processor in superLFM mode? As in, it actually downclocks the FSB to half its normal value?

I keep getting system hangs if I enable it, even with CPUgenie. I'm starting to think Dell has something to do with it, i.e. by not supporting it through the BIOS. Which would be rather idiotic since it means disabling a feature that can preserve more battery life which is not unimportant for a laptop, at least in my mind. As a matter of fact, CPUgenie doesn't list superLFM as a p-state found in the ACPI table in my BIOS.

 

How exactly would you do that? If I don't set my min. & max. proc. power states to 100% in seven, I get weird VID spikes even tough RMClock's profile is in control. I get this with OS power management integration enabled or disabled.

 

The IDA spikes seem to go away during CPU processing and come back after prolonged idle. Maybe you can do an experinent for everybody and see if there is percentage somewhere below 100% that still gives top FID without IDA blips.

On the surface, it might seem that SLFM would save battery. It doesn't. Because two FID's runing at the same voltage take the same watt-seconds of power computing a digital job before the going back to idle (no load). Twice the power for half the time is a wash. SLFM is a joke for battery purposes. Use Standby instead.

 

Not sure if its been asked but I don't wanna go through 600 pages. How do I get the program to start without win7 asking me to click yes everytime I boot up?

 

Wow great explanation!

So I've now undervolted my Vostro 1500, but am worried because I dropped the voltage so much, even though no BSOD yet.

Do the following numbers seem crazy low?

6.0x 0.9500V
7.0x 0.9625V
8.0x 0.9625V
9.0x 0.9750V.

Also, I didn't mess with the SuperLFM and IDA voltages. Can I lower those to further drop heating?

Thanks!

 

Quoting myself (used in both Vista and my new Win7 Laptops)

 

Not crazy at all! Remember it's all good if you don't get any BSOD's... And even if you do, nothing will be permanently damaged (other than maybe files you are editing at the time of the BSOD). Your worries are unnecessary since undervolting is completely safe - hardware wise. (Please correct me if i'm wrong here).

My values:

9.0x 0.9625V
8.0x 0.9375V
7.0x 0.9000V
6.0x 0.8750V

This is on a Dell SXPS 1640 with 2,4ghz C2D.

Other than that, I enabled SuperLFM (6.0x) and set it to 0.8750V as well, but not sure if it actually makes a difference.

Good job tho =)

 

Not crazy at all. My C2D 7200 uses those exact voltages (starting with 11x), and my temps dropped from 93C+ to about 74C at Full Load. I have been running it now for a couple of weeks at full load with no BSOD and no rise in temperature.

 

Tried with CPUGenie, and here I got:



Quite impressive, I guess?

Forgot to say, my CPU is P8600 (you can guess it, right?), and my laptop is Thinkpad T400.

 

Is there anyone with a Penryn CPU got a Vcore lower 0.925v?

 

can some1 plz help me i don't have access to those options like those in the screen shot(rm) like advanced cpu settings and most of my options are greyed out i have admin privileges.

My laptop is an acer aspire 5536g with a AMD Athlon 64 X2 QL-64 running on windows 7 Ultimate 32 bit.

I really need help cause my cpu keeps over heating and than throttling making me lag bad.

 

try this http://forum.notebookreview.com/showthread.php?t=390696
You might be able to go lower but that all depends on your CPU

 

An interesting tutorial, but unfortunately, I'm not a hardcore user. I feel much safer to undervolt with software.

So 0.925v might be the lowest VID of a Penryn CPU. CPUGenie shows that it can unlock lower Vcore, but nothing's done in reality. Even when I get to 0.85v with it, CPU-Z and Coretemp still show 0.925v.

Anyway, thank you for the link

 

Try k10stat and see if it works.

 

Hi Scott, thanks for the throttlestop idea. I am a newb so looking for the easiest, "foolproof" way to undervolt. Trying to reduce my CPU temps and fan noise. I have W7 and T5600 1.83Ghz Intel Core 2 (on an old T60). Just a couple follow-up questions:

1) If I run TS, do I no longer need to run RM Clock?
2) In TS, which boxes should I check by default? I noticed you can change "set multiplier" and "voltage ID" but don't want to fool with these until I have some guidance. I am not sure how to "Set single CPU voltage and FID" as you mentioned above....

Any ideas? Thanks!

 

1. Yes, you should pick one, not both.

2. With TS, right now all you can do is pick one multi and one voltage.
So speedstep will be disabled. Just use your highest multi + lowest VID at which that FID is stable.

 

Hi, doing this first time... I have HP HDX16 running very hot, got 100C.

Now, I did go through guide but have questions...

1)Under CPU info, it shows Intel Core 2, CPU CORE: unknown, revision unknown
is this a problem?

2)Under performance on demand, below P-state, I have only 2 states. 6X and 7X
in your example you had more, is this program not detecting my laptop correctly?

thx!

 

Anyway to undervolt a Core i5?

 

CPUGenie

 

Have anyone tried it and succeed ?

 

It seems that I've go it wrong. From their official website ( here) it seems that it doesn't support i3/5/7... though I'm sure I've read a thread where someone claimed using it successfully on his i7

 

HALF MULTIPLIERS FOR RMCLOCK 2.35

first of all i havent followed this thread for awhile since i got my undervolting under control never came back, so if this has been addressed before.. then oh well!

so some of us have half multipliers (mine is 9.5x), but RMCLOCK 2.35 only allows u to undervolt whole multipliers. therefore u miss out on ur last 0.5x of cpu juice. i been fine with having 2.4ghz rather than my full potential 2.53ghz for over a year. however, i found this thread from google and i am now running at "10x" (cpu core clock says 2.527ghz so its really just 9.5x). remember to restart the undervolting process for the last multiplier or u will get blue screen!

here is the forum i found:
http://www.overclock.net/laptops-notebook-computers/543246-rmclock-half-multipliers-i-did.html

EDIT: To add the final half multiplier rather than replacing the multiplier. simply copy and paste the final four binary pairs at the end and change the previous multipliers back to original value and you will have all the original multipliers with one extra multiplier that is set at the voltage you have undervolted it to. on my rmclock, the VID drop down is screwed up and won't show my voltage, but on the monitor and the custom p-state drop down... my voltage is where i set it.

 

Hi im desperatly trying to undervolt my p8400 cpu but every program i use does not show the correct temperture it just stays at a random figure and doesnt move when i use orthos,

does anybody else have this problem, or does anybody know of a program that will correctly read my tempertures?

Thanks!!

 

Have u tried CPUID Hardware Monitor... it usually shows right temperature... also after undervolting , u have to tweak RM clock to work as u have half multipliers..... here is a link for u to have full cpu speed...

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

Hope this works.

P.S : Another one u can use is speedfan , realtemp...

 

I have windows 7 64bit and installed rmclock using the driver attached with the guide for 64bit users. After hours of stability tests, everything is fine. However, after a period of time of normal tasks, i get instant turn off (no warning, just turned off instantly, no restart!?!). The event log shows it as event id: 41, source: kernel-power. Anyone got this problem?