The "Undervolting" Guide

hi, scott.ager, thanks,
it is just difficult to follow a 430-page thread even using search function.
I had seen in some other forum or article that V is locked by the manufacturer but i wasn't sure and i hoped that of that was the case someone could help with that too.

It's just frustrating. I mean it's not just the lowest multiplier that can be run fine with the lowest voltage available, there are other 4! and i have lowered the bigger multipliers an average of 15% in voltage...


So, there is no way to unlock this lower limit set by the manufacturer?

 

There are no Nobel prizes for undervolting. You've gotten all the gains allowed by your notebook model. 15% is a good result based on hundreds I've seen. Why don't you study a programming language instead and write free software for all of us

Oh yeah, BTW, your CPU only uses the top and bottom FIDs you checked. The rest aren't used. Hope you didn't waste any time undervolting them........

 

1. No, I'm not surprised. I read their posts.

2. The UV "game" is rigged Some notebook makers lock out low voltages to stop BSODs before they can happen.

3. Unlike in overclocking, there's no prize for lowest voltage here. UV is just a tool. Set it and forget it

 

ok, scott, i was just asking, relax, thanx again anyway.

p.s. i did. i did find the lowest voltage for all fids,
once manually via rmclock, and once using cpugenie... oh well...

p.s2 "learn a laguage?" i am not that good at computers.
i ll stick to what i do, that being a biology student, at least there is a nobel prize for that. lol!


but i ll do my part, i ll post on my models owners lounge thead a table with the vids for others to see.

 

Looking for the RTCore64.sys driver that is supposed to make the thing run under Vista 64. I have Win 7 x64 (love it...)
I found a few rapidshare links to RTCore64.sys, but they are all not working.
Why can't this RTCore64 be on the RT core download page.
http://cpu.rightmark.org/download.shtml
anoying.

Cheers
M.

 

The first page on this thread is really useful.

http://forum.notebookreview.com/attachment.php?attachmentid=24096&d=1221194853

 

Will this work in Linux?

Sorry if this has already been asked.

Also someone told me I can use this to overclock my laptop CPU without it being as hot.

True?

Would liek to go from 2.53ghz to 2.8.

 

It will not work in Linux, unless you can run it in something like WINE.

Not true. You can overclock, then use RMClock to undervolt. The undervolting will get the overclocked CPU running as cool (least voltage) as possible without external cooling.

 

Your Aussie mate, flipfire, put it in the UV manual knowing that it would be the last place people would look for it

"Here is the link for the signed RTCore64.sys driver: (Virus free and tested)
www.flipfire.net/download/rtcore64.zip
Download and extract it to your RMclock folder and overwrite the old driver"

 

i installed rmclock and it automatically installed into the Program Files (x86) folder. is this correct? also, i can't seem to stress my computer correctly. running windos 7 x64 professional. the cpu states half of my clock speed at stress only. been doing this for years on windows vista but i JUST upgraded to windows 7 and am helpless.

 

Use the voltages you had back with Vista. I'm guessing your CPU chip is still the same?

 

Running an Intel CPU? Need to stress it? No need to be stressed out...

use:
http://downloads.guru3d.com/IntelBurnTest-v1.6-download-2047.html

(at your own risk, of course).

 

furthermore, i'm running the SAME voltages i had with vista. yea everything's the same. just some clarification, again. hehe..

1) installed into Program Files(x86) folder <----- is this correct? or should it install in Program Files folder?
2) as 1), RMClock is now not listed in Programs listing (to be uninstalled via Control Panel), so what do I do now?
3) I downloaded the RTCore64.sys and overwrote the default version as per installation (should I NOT have overwritten it?

*** Could anyone guide me on what you did exactly to get this running smoothly on Windows 7 x64 Professional?

As of now, I believe no matter what the case, I'm running at near Power Saving speed... my CPU's T9300 and it keeps running at 6.0X, which is the lowest! I turned off RMClock and used the windows power settings to test out. Even at high performance, running a few programs, it's at 6.0X!

 

1. Doesn't matter.
2. Use the registry wipe file that comes with RMC. Russian uninstall method.
3. The signed x64 driver is correct. The original just won't work.
4. I've found that RMC should only set voltages and the W7 power plan must do the rest. That means no OS integration.
5. Twiddle the knobs like everyone else. RMC was written before W7 existed.

 

ok thanks for the reply.

i noticed that after uninstalling RMClock, my CPU becomes dynamic again, ie. varies in speed in accordance of requirements. however, once i install RMClock, even when i stress-test, the speed fixes at 1200MHz, which is 6.0X for my processor, ie. slowwwwwww. Have any of you guys encountered this phenomenon?

OK solved the problem. i turned on thermal monitoring 1 and 2. turned them off and now it works like a charm. it was due to thermal monitor 2 which locked the speed to 6.0X.

 

how to do undervolting on ubuntu?

 

Is it possible to undervolt the i7 with RMclock ??

 

i installed rmclock on windows 7 64bit downloaded the 64bit driver but will not recognize my proccessor just seas n/a were ever it talkes about the proccessor?

 

It would help a lot to post the name/model of your CPU

 

AMD Turion 64 X2 in a gateway NV52

 

Try this. http://aspiregemstone.blogspot.com/2009/06/k10stat-amd-griffin-processor.html

A search of this thread for "Turion 64" yielded 48 hits leading to the conclusion that RMC won't work on your CPU.

 

Read the intire thread didnt see anything that talked about what you said just talked about K10stat program

 

I've never owned an AMD processor so I'm no expert. What I read here is that RMC works with AMDs up until the newest models. If RMC doesn't work right with your chip then you have a choice of CPUGenie or K10STAT. http://www.mediafire.com/?dywh3zhmk41

My search popped up #3005 by miner.

 

sounds good ill try thoughs

 

Just reinstalled 7 because my ssd came yesterday, and I am noticing that with Professional, RM Clock did not undervolt to .8750V all the time. In Vista HP and now my 7 HP upgrade from Vista, at 7.0X it reads .8750 unlike .9000 or higher in Professional.

 

I just did my G50Vt-X1. Got it down to .9500 V and underclocked it to 6.0X. It was from 52C down to 40C at full load. Shaved 12C!!! Thanks to this thread!!!

 

This Post needs to be in the first post, or have a link to it.

 

I agree with you

 

My vote for first post is: "What Is Search This Thread ?" or "How I Forgot To Use The Library".

A search for "UAC" turned up 43 hits in 0.48 seconds with how-to recaps by Ferretwulf in the first 10.

 

????

Someone first reading this thread is not going to know where to find the post or what to search to look for it.

 

Then, perhaps the Manual should have a short preface called "Note to Vista and W7 users:".

Key points: UAC and signed x64 driver.

Just those will cut a ton of inquiries.

 

Are there lower voltage options possible then just .925?
Reason why I ask is that my mobile T6400 is already at it's lowest even at highest multi, and it happily churns away at orthos for hours on those voltages.
If it could go lower I coul save even more energy/battery,

 

I just wanted to add my 2c on this. I am writing this mostly from memory from back in school and while many things in the underlying technology may have changed, it should still hold true.

I currently have a Pentium M (Dothan) 1.6GHz CPU and I have my voltages ranging from 0.732V to 1.02V. My lowest multiplier is 6x for 600Mhz. I was able to run at 0.700 for more than 3 years but as my components have now aged (laptop is >5 years old now) I bumped up every voltage an extra 0.032.

All (Si) transistors have a minimum switching voltage of 0.7. In this regard, you cannot set a voltage of 0.5 and expect anything (good) to happen. If you set the voltage at or above 0.7 in theory (for a single transistor) you should be able to switch it from 1 to 0 (open/shorted, on/off). However, it does not work like this in complex systems as CPUs have a huge number of transistors and curtain physical processes related to the nature of the Si transistors and the inner-circuitry make the CPU require somewhat higher "supply" voltage to operate error free.

While according to people in this thread that have run numerous benchmarks say undervoltaging does not show any changes in speed, it is my understanding it should actually lower the speed. The reasoning behind it should be simple (in my understanding) - the less current that flows through the base the longer it takes for the transistor to go from open to closed. This seems to me is the same reason why you need to increase voltage when you overclock - to allow the switching process to happen fast enough.

I think that I may be somewhat mistaken, as the fact the frequency stays the same means that the speed of the switching if high enough... Would really like to read what others have to say about it.

As I am waiting on my new laptop to make it from China and it comes with W7Pro I will have to start looking for alternatives to Notebook Hardware Control ( www.pbus-167.com). It seems to be similar to the new program CPUGenie, as it has an easy to use GUI, simple built-in error checking when you change voltages and such but it doen not work on Vista and Win7. I would recommend it (Notebook Hardware Control) instead of RMClock if you are using Windows XP.

 

Glad to hear from someone interested in the basics. One small but very important correction. It's true that the volume flow of electrons per second decreases with decreasing voltage as per ohms law. But, in a digital circuit, like a fire brigade, you only need to pass a full bucket of electrons to the next gate within the clock cycle for the circuit to work. Partially full buckets from too low a pressure (voltage) count as an empty and the brigade comes to a halt.

Overclocking allows less time for a full bucket of electrons to fill up at a gate and so it needs more pressure (voltage) to make it happen. Again, partly filled buckets count as empty or zero, not 1, and presto, BSOD.

Re: CPUGenie. Lots of us like it but it still has stability problems in sticking to setpoint voltages.

 

Perfect Everyone gets to answer their own questions.

Your notebook maker locked your lowest voltage because they don't trust you

 

Don't see what is so funny about this. It was a serious question. It's at the lowest possible option in rmclock.

 

I don't know your OS but if using Win XP try Notebook Hardware Control. It allows full voltage control (0.7V and up) on my laptop at least. I really like the program although the last few revisions started utilizing MS .NET framework and this (in my subjective view) made the program more bloated and slow.

I made my post a bit confusing after the edit about NHC. To clarify - I have not tried CPUGenie as I am still on XP and for me Notebook Hardware Control is working great. As my new T400 comes (it is in customs now) I will have to look for an alternative only because NHC don't seem to work on Win7. A new version was supposed to be out for running on Vista (and Win7 possibly) but it never happened.

Your point on the "electrons brigade" is absolutely correct. In digital circuity you only need the transistor to "switch". As long as the voltage is enough to make the process happen in the allotted time it don't matter if it happened real quick or at the last moment.

 

You should point out that the last NHC version released was 2.5 years ago and it requires a custom ACPI interface module for each computer model-mainboard. If you have an older, popular model, you're in luck. If you have a newer or less popular rig, you're S.O.L.

Their website has dozens of appeals for modules with no answers. I tried my hand at ACPI programming but gave up. My Fujitsu sensors are embedded and locked up anyway. Good luck

 

Apply the RMC tweaks registry file to unlock the range of voltages that RMC displays whille setting your VIDs. Caution: just because you choose a voltage in RMClock does not guarantee that it will be applied. Your mobo bios selects the lowest allowable voltage and RMC cannot override that value.

BTW, did you correctly select "Mobile" at the bottom of "Advanced CPU Settings" ?

 

For some reason rightmark is seeing all of my voltages as 1.25 for multipliers 6-10. I looked up the specs on my processor and it says voltages 1.075-1.25.

I am currently running orthos with the 10x multiplier set at 1.075 and it hasn't failed yet. I am going to reboot though and see if rightmark will read things differently.

I can tell you this however, in a short term test my stock cpu with a 1.25 voltage hit 80 degrees. Now with a 1.075 voltage I am hitting 65 degrees. That is a 15 degree drop in max temps which is absolutely amazing, however I don't know if this is 100% stable yet or not. I will reboot, and see if rightmark reads anything differently and go from there.

Thanks again for this thread, this will be a HUGE help!

 

In the 'Profiles' main window, did you click the 'Defaults' button? Doing that loads the default settings from the BIOS, then start undervolting.

 

If it weren't applied it wouldn't show in cpu-z now would it.
As cpu-z shows .925 volts being given tells me it is being applied. And yes, I selected mobile.

 

I still don't understand, cpu-z or no cpu-z. You can't select a voltage choice that is not being offered to you. Try CPUGenie. It usually offers up a wider range to select from. If that doesn't work, then you're just plain locked out of going lower than 0.925

 

I never knew that one. Thanks, cowboy

 

Has anyone undervolted a SU7300 here?
I can't choose different multipliers in RMClock, only 6.0x is shown. (using win7 64)

 

Are you using the signed 64 bit driver and the correct startup procedure for use with UAC ?

 

Driver is installed. Don't know what the correct startup procedure means. UAC is turned off and i start the program as administrator.

I actually haven't seen anyone reporting about undervolting a SU7300, so who knows?

 

Hey guys;
Not hure if it'll show my signature but I'm running a T7250 on Windows 7 Home Premium 64-bit. I was wondering if I should undervolt my processor. I upgraded from Vista Home Premium 32 and I used to run it undervolted then. My temperatures right now are idle at 40C, while I stressed it to max load and after 20 minutes it was up at 81C however never exceeding 83C. During normal usage like chatting, music and browsing it's normally around 55C. I'm wondering if the max load temperature registered is alright? or is it hazardous? When I undervolted in Visa my temperatures reached a Max of 67C at full load, and would hang around 49C with an idle at 43C. I had the SuperLFM disabled.

Any help on this would be appreciated.
I stopped undervolting because after a while on I kept getting BSODs on Windows 7, but they seemed to all be video drivers related, however after troubleshooting everything, with every possible driver, this still happened, so I formatted, no undervolt and so far, smooth as pie, except for that one concern, I mentioned.

Help please.

 

Are we going to be able to undervolt Arrandale CPUs at all?

 

Yeah, yours may not work since Intel already cranked it down to a 10W CPU. Who knows is right.