The "Undervolting" Guide

When RMClock is running, I notice a random delay when typing in forums, what is the problem? Is it because I'm using SuperLMF?

 

I tried to uncheck everything, but the 7x voltages are the same (1.1250 first, so i cannot increase it).

 

so all of your voltages are 1.1250?

 

Can anyone help me on this?

 

can you try turning UAC on?

 

I think Flipfire posted the special 64bit driver on this thread a long time ago. The stock driver has permission problems or Vista problems or something along those lines. It's here, just search.

 

Looks like I was lucky and got a pretty good CPU.

Lenovo T61 4:3 screen laptop. SantaRosa 2.0ghz T7300 G0 revision CPU. Default max voltage is 1.1875v at 2.0ghz as reported by all apps. Oddly others report their T7300 CPU being 1.25v....mine must be special

800mhz - SuperLFM - .8500v - (default)

1.2ghz - 6.0x - .8500v - stable (lowest possible voltage)

1.4ghz - 7.0x - .8500v - stable (lowest possible voltage)

1.6ghz - 8.0x - .8500v - crashed
1.6ghz - 8.0x - .8625v - stable

1.8ghz - 9.0x - .8875v - crashed
1.8ghz - 9.0x - .9000v - stable

2.0ghz - 10.0x - .9250v - errors
2.0ghz - 10.0x - .9375v - stable

Max temp has dropped by 14c

 

Just realized the RMClock eats a few % of CPU cycles when its running due to the useless graphs it runs in the monitoring section. Anyone know how to turn off these useless monitoring graphs???

 

Hi all,

I was happily working on lowering my voltage, (my highest temp was 86c!)
So, my 10x was set at 1.250, and I lowered it to 1.050 with no trouble. However, for some reason, I can no longer go any lower, on ANY muliplier. All of them are at 1.050. When I click on the drop down menu, I can only go higher, no lower.

Any advice? Please speak slowly, as I'm new to this! :lol:

On a good note, my temperature is lower, (about 75 or 76c), but I wanted to see if I could get my voltage lower, as i never found my crash point.

Thanks!

 

Question about the 'CPU Load' in RMC. It always hovers around 40 to 50%...is normal?

I'm on Win 7 x64.

Pic added...

 

This is too bad. Not sure if its worth keeping RMClock installed if it uses all these CPU cycles... What were they thinking???

 

Theres no way to lower it unforunately, Intel has hard locked the minimum voltages that can be set.

The RMC monitor tab monitors the CPU utilization of the current/active multiplier, not the total CPU load. So say 58% of your 6x (idle) which is 1200mhz

 

Does not matter. The point is this RMClock program eats up alot of CPU cycles and there is nothing you can do about it. The monitoring graphs it runs are useless and there is no way to turn them off to stop the program from using so many CPU cycles. A program like this needs to be "lite" since its running all the time.

It actually sucks that it uses so much CPU power for those useless monitor graphs. I uninstalled it

 

Back in Russia, in 2007, they were thinking a couple of things:

1. RMC works great on XP and Vista using the latest even-multliplier dual core CPUs.

2. We're not making a ruble (dime?) on this freeware. We quit.

We encouraged them to become capitalists. Be careful what you ask for ................

 

@ 2GHz:
Voltage drop is approx. 21%
Power drop is approx. 38%

Now, don't be a slacker. Leave it on 2 GHz all the time. Standby (S3) when not active.

 

Well, im undervolting my laptop and i reached the minimum voltage of my laptop but after i reached the minimum voltage i used one file to unlock more vids, so i reduce more but i think the program has an error
In this picture, the CPU is under stress and the VID is 1.1000V


But it shouldn't be that VID because i reduced the VID in the main profile as you can see here


My doubt is which VID is the processor using
Sorry for the english, im portuguese

 

Heyya guys,

I am new to undervolting and i will undervolt my lappy soon.i am worried a little. i want to ask that have you ever has any problems with undervolting?

Is it guaranteed that undervolting is safe?i want to be sure cuz i you know

And is undervolting fully stable when udjested?

 

Undervolting is safe in the sense that it would be extremely hard to damage anything hardware wise with it.

That being said, you could have instability issues or a couple of BSODs. Nothing serious but if done improperly, it could be a real hassle.

 

Is it possible to undervolt sager NP5797 in bios?
Thanks

 

Hi folks!

Just want to thank for this great thread, has a huge different on my laptop, Zepto 6224W. I had huge heat problem with it, I almost burned my fingers on my laptop while writing on it during summer. Under the cold winter here in sweden it is kinda nice warming on the laptop after a walk in -15 degrees. But during the summer it is not as fun. In the beginning I suspected the graphic card (8600M GT) but no change in temperature when I underclocked it as much as possible.

Then I found this thread! I tested the heat of my cpu: 95 degrees! Then I began undervolting it, from 1.1375V to 0.9500V! From maximun to minimum voltage. Ran the test for 9 hours while at work without any error or instability. And I went down to 74 degrees, 20 degrees less! It is even cooler then my GPU now!

Edit: I tried the difference with wPrime. With RMClock off I got 32mb done in 40 second, before and after I tried with RMClock on. However, when I start RMClock this time goes up to 46 second with maxed and minimized volt. Why is this?

 

The only issue for everyone is the high CPU load RMClock poses on your system due to the useless monitoring graphs it has running which you cant turn off.

 

Thank u for this guide... will help when i upgrade my processor!

 

I have been in the process of undervolting my Dell Studio 1555 (specs in sig) and I have a few questions.

I have gotten the 10x down to 1.0125V with only 1.0V left go. I have been changing the voltage and then testing for 45 min as per the guide. Should I continue doing this? I have read some of this thread and I have seen mentions of 2, 3 or more hours of stress testing. Do I do this now or after I have gone through and am done with 6x and above?

How accurate are the temperate measurements in HWMonitor? At the moment it is saying my core 0 and 1 are 57 C whereas RightMark is saying 41.5 C as the "Core Temp." Are these the same measurements?

 

true.

It constantly uses 4% of my CPU

 

Dear btuger - Please use thumbnails for your graphics. See the Attachments button in Additional Options.
Big graphics makes people scroll way left and right just to read the text.

 

Actually, I wish new people would spend more time watching the graphs while they work instead of rushing in to post their voltage "scores". If one really wants to maximize the opposing forces of battery vs.performance, one needs a better understanding of what clock speed and load really are. Geez, how many frequencies does your CPU really use and for how long in everyday computing? Microsoft could have made a ton selling a "MS-Orthos" right here.

 

Yup. I uninstalled RMClock. Not worth having my CPU running at 4% when it should be at 0-1% tops for such a simple app.

 

I'm actually thinking of not using it too.

I wonder if anyone has done any tests to see the effects on battery life?

 

I'm very curious as to what percent load you run at without RMC. I mean, the 30+ processes and device drivers running in the background aren't free...

Also, it's like the Heisenberg Uncertainty Principal: the very act of measuring something changes the outcome

 

If you plan on using each multiplier, I'd continue testing each one as you did for 10x.

I know when I started undervolting a few weekends ago, I was anxious to see results. So I did 10x, tested for about 15 minutes, and auto-adjusted the rest.

Monday morning, after about an hour at work, the notebook blue-screened without warning. Oops. Sure glad I saved those couple hours the day before....

I went back and redid it carefully and haven't had problems since.

 

Anybody else have this problem after undervolting?

Under "CPU Info" tab, there's the "Current VID" which should be whatever multiplier you're on right now and what you undervolted it to. It's bothering me however, that under "Startup VID" it says 1.000V for 6.0x when I undervolted it to 0.8750V

Furthermore, for "Maximal VID" it shows 1.1250V which is the largest possible for the 9.0x but I also undervolted that to 0.9250V. "Minimal" is at 0.8750V, which makes sense to me. Whenever 9.0x is under "Current", however, it still shows only 0.9250V instead of the maximal 1.1250V. That goes for all multipliers under "Current", they'll all use the undervolted settings.

I also went to check the monitors and realized that the VID spikes up to 1.000V quite often with rare cases of 1.1250V when idling. I have no idea why this is happening, anybody else has this problem with the P8600?

 

Yeah, RMC pulls some of it's numbers off the defaults for your chip. I think they "lifted" that module from cpu-z or somebody else. Ignore it. Check your monitoring window and believe that. Turn off IDA and SLFM to get rid of twitchy spikes.

 

Thanks, and I checked the monitors and the VID does spike up to 1.000V and even 1.1250V at times. IDA and SLFM are turned off as well. I don't really mind as long as the voltage spikes aren't harming the cpu though.

 

If you're on Vista or W7, read back aways. Somebody had a fix for that (phase 2, after unchecking IDA and SLFM ). Seems the newer OS power plans are more forceful in fighting with RMC over control.

 

Probably

 

The cpu usage hovers at 0-1% (while not running RMclock).
when I'v got RMclock running, CPU usage hovers around 5-7%.

explain that.

I'v decided to not use RMclock. Now I just overclock my cpu 33% with stock volts.

 

For those of you complaining about CPU usage by RMClock. You can turn down that CPU utilized quite a bit by changing the monitoring interval (how often the graphs refresh, etc) in the RMClock_Tweaks.reg file.

1. Open the RMClock folder and find the RMClock_Tweaks.reg file
2. Open it in Notepad
3. Change the RefreshInterval to "RefreshInterval"=dword:0000EA60
4. Close Notepad and save changes
5. Double click the .reg file and say Yes to add it to registry

Then relaunch RMClock and you'll see that the graphs update every minute now instead of every second!

 

I still get ~1% load with RMC plus 30 processes on XP. I know it demands realtime priority which probably gets it into trouble with newer OS's. Too bad, but the developers couldn't afford a crystal ball and newer code by writing freeware

 

Thank you, sir
I was meaning to do that and you saved me a lot of time.

 

So in xp it's fine. cool, I guess I will use my XP HDD when I need max battery life.

 

@scott_ager

Why don't you need to use SLMF? Is it bad and not recommended?

So all that's recommended to check is the lowest and highest multiplier? Or is it SLMF, lowest, and highest multiplier?

Where is the post indicating the fix for Vista/Win7 having voltage spikes? I can't seem to find it in the previous posts

Lastly, for the P series (low power) C2D's are they pre-undervolted version of the T series which explains why you can just undervolt them so much without getting errors? I have a P9500 and the 9.5x default's VID is 1.075, all I can do is undervolt it to 1.0375.

Thanks.

 

For me, using SLFM is about the same as standby. What's the point of slowing down your computer to where it doesn't run well, especially with Vista and W7....

It's a free country and I'm not going to tell anyone how to use their own computer. IMHO, undervolting is essential since the mfrs. won't do it and speedswitching is a waste of time unless your notebook still overheats.

There have been several references to tuning Vista and W7 for RMC in the last 2 or 3 weeks. I didn't make a note of who or where because I have no intention of "upgrading" to any OS that uses 3 - 4 times the memory and 10 times the HD space.

I'm pretty sure the low power chips use a special design and because their selling point is low power, I'm also pretty sure Intel squeezed down the voltage range to help get there.

 

Ok, so you mean the stock VIDs of the low power chips are lower than the stock VIDs of the standard power chips?

 

This is primarily a problem of Windows. Windows 7 is supposed to fix it but at the moment both XP and Vista(also earlier windows) switches threads around and keeps the processor awake, preventing power management to work effectively. The primary reason they do this is apparently the multi-tasking performance on single core CPUs are better(yea you read right, multi-thread on single cores), but now that we are in the multi-core era, that's a problem.

AMD's C&Q not working well, and SpeedStep having problems ALL have to do with Windows.

Ideally, the point of technologies like SuperLFM and C-states aren't solely to lower power when the CPU isn't doing anything. See, CPUs aren't loaded to 100% even on applications that seem to do it. It's like how with humans you can find few mins/even few seconds of break even when doing intensive work.

An easy example is playing back video. A relatively CPU intensive operation, but it isn't loading the CPU at 100% all the time. The CPU will do whatever it is necessary to get the frame, and send it to the GPU to render it. Between the milliseconds it has till the next frame, the CPU can go to low power states to save power during intensive applications.

Summary:
-CPU power states are used to save power not only in idle, but on load
-Windows currently messes up with this operation, which means it doesn't work as well in theory
-Windows 7 is supposed to fix it, but we'll have to see if it really works

 

So with that, what P-states you suggest I use?

 

No, 1.1250 is the highest voltage for every FID, but trying to apply lower values, I got crashes in the next seconds.

 

Here it is, from two weeks ago. Nobody wants to go to the library themselves anymore.Thanks Chumley.

 

Thanks

Are Speedswitching and Speedstepping the same thing?

 

Agreed. Just as we carry around lots of vestigial DNA from ancient species, notebooks are burdened with leftover early chip architecture and operating systems written for desktops with unlimited power supplies and cooling capacity. Too bad the W7 "solution" is more bloated than a prize hog at the Cloverdale country fair. One step up, two steps down

The Fraser Arms(chair) Philosopher

 

So techincally, I should just disable OS Power Management Integration, and use one of Vista's power plans? Any of the three will do?