The "Undervolting" Guide

i haf a question.. im done undervolting, but when my laptop is idle, the cpu load has spikes between 30+% - 70000%. and the FID and VID also has spikes from 6-9 and 0.9v-1.2v at times.. is this normal?

 

hey scott, is this ur default setup for RMclock for ur laptop? cuz i was reading bout ur posts bout how speedstep is an amusement.. and i'd like to try doing without it, but how do i go about keeping my lappie at max clock? using these settings of urs?

 

Great. Yes, I only use 1 FID like my avatar says. Somehow I get the feeling that people are confusing speed stepping as being part of the undervolting process. They are in fact, two completely different things. Undervolting makes stepping unnecessary. When you cut the factory voltage to your CPU (and test it for reliability) you eliminate the need to step down to control overheating. How would you feel about buying a car that had to slow down to 30 on the freeway every ten miles to keep from overheating ? Welcome to "Laptop World".

Anyway, I just run max profile all day long, on-battery included. If you have a utility in your desktop tray that shows CPU load as you work, you'll see that it spends most of time loafing like bum. Maybe I'm just just getting cranky as I get older but IMHO, gamers should use desktops. They're more powerful and a lot cheaper. Ditto for video fans. On the other hand, if you run Orthos all day, well, we have hospitals for them too

 

when u use one FID i.e. the highest, there's no shortening of ur batt life? and and, u say to tick IDA.. but isnt IDA using one core while the other rests? shld i be ticking my highest multiplier instead? and.. u mean running Orthos too much is bad?! but i need to run like 45mins each voltage to test... is tt alot? lol.
and one more thing.. if i use only one FID, does it mean i only need to test that FID for stability?

 

and also.. my CPU load has spikes tt goes up to 100000% on RMcclock cpu info.. why is tt?!

 

Turn off (uncheck) IDA in adv. CPU settings. Click on thumbnail.

 

Battery life depends on how much you use your notebook. That's load.

No, I said UNtick IDA.

Only select the multipliers you're going to use.

5 mnutes of Orthos told me all I needed to know. Just test the ones you're going to use. You can guesstimate the answer for the others.

 

so i juz untick everything like ur thumbnail? including the thermal monitor 1? cuz in the default there'r alot more other ticks.. they'r useless?

 

Not useless, just not necessary for this one scenario.

 

Anyone, please?

 

How far down have you gotten your P8400's?

 

It's totally up to you. Check the IDA box if you want to run at your CPU's rated speed (RMClock rounds down 1/2 point in actual speed but indicates down a full point). As far as SLFM goes, well, it's your machine. Let me know if you find something it's good for.

 

Sorry but I dont understand CPU rated speed. What is recommended? Am I suppose to check those boxes when I am done undervolting to put the changes into effect? Will either box help my computer run cooler without compromising performance?

 

Dell Studio XPS 1640 Intel Core 2 Duo P8700 @ 2.53GHz - Your signature says you have a 2.53 GHz speed, Intel P8700 CPU. I'd say that your top clock speed is 2.53 Giga Hertz (GHz) which is equal to 2530 Mega Hertz (MHz). Your chip was designed to run at that speed or slower.

 

Rather late response but I'll post it anyway. There is a possibility to do this. Use task scheduler to run RMClock and select "even when not logged in". This runs RMClock in the background, no icons, only visible in task manager when you view processes (click show all ...). I'm using vista 32 and I have no icons showing. As far as I know this is also the only option to run RMClock automatically at startup without UAC prompt & extra running software on a limited user account.
(Not that I'm very knowledgeable about this but if someone has another way, please let me know or post it here)

Now the tricky part. I just managed to set this up on a (to my memory) fresh install * without problems, but on a previous install I got the error message "this user has no right to logon as a batch job" or something similar when selecting "even when not logged on" in task scheduler. I managed to overcome this by setting this right for my admin account using a utility called ntrights which can be found in rktools.exe, an MS tool kit not intended for (but nevertheless working) on vista. Just ignore the message when installing.

Should work fine afterwards, but remember that I have no idea whether this has a negative influence on your computer security. As a matter of fact I do not fully understand what "logon as batch job" does, so if anyone wants to share his/her knowledge ...
And finally, again, if anyone has a better way of running RMClock automatically on a limited account, would like to know this too. Scheduling a task with "only when user loggs on" doesn't work for me.

* actually it's been a while but I haven't messed with RMClock yet on this install

EDIT: As an extra, I see a lot of posts from people with voltage spikes, maybe a FAQ entry can be included for this problem or something. I had the same problem but managed to fix it by setting my power profile from "balanced" to "power saving". This causes the min. & max. CPU settings to be equal (100% surprisingly enough, I thought power saving would be more power saving than "balanced") which solves interference with RMClock. >> Source. scroll down to iii) bug #3

 

1313 - I followed your link down Memory Lane to a year ago on this thread. I still can't see that anyone has provided physical proof that speed switching results in energy savings for normal computer use. Do you know of any "Orthos" type utility that will give a fixed percent load to the cpu and then provide completion time for a standarized task?. Right now all I see is 100% load for a user defined time interval. That's to check for CPU errors. Rather than buy a multimeter, take my battery out, and slice the wires to my AC charger, I'd just as soon use an enery-measuring software program to see if speed switching really increases battery life under real use conditions. I secretly suspect that it's just plain fun. Was for me, for a while.......

Cheers,

 

Scott,

Euhm, I presume you read the entire post I linked to (that's actually in another thread) and are commenting on one of the things handled there.
My previous post hasn't got anything to do with speed switching vs energy savings though. It's about people having VID spikes in RMClock's monitoring graph. I noticed this was an often returning question and encountered it myself, so thought I'd mention a possible fix. I edited my post to be more specific.

I havent' actually read the entire post I linked to, it's on my "to-do-list", as are many things.

 

1313 - Sorry, memory lane meant a long stroll thru last years hot topics. I did actually do something relevant to spiking: I turned off IDA after selecting an IDA type FID (the highest one) and that was the last of spiking. And, I got the half muliplier speed.

Cheers,

 

Hi, I am having trouble installing the RMclock v 2.25 I read in this topic that I need to install a driver to allow this progam to work on a 64-bit OS. However I am having trouble with the installation. I am not sure how to extract the driver to the RMclock folder. Otherwise I keep getting an error message saying that I cannot install RMclock.

 

My CPU is T7700 2.4GHz and the max multiplier is 12.0x.
I always limit my CPU to 1.6GHz with multiplier 8.0x to keep my laptop cool.

My laptop already 1.5 years old and I noticed that that the lowest stable VID is increasing as time past by.

For example, when first time I undervolting my CPU, I set FID 8.0x with 0.8500 (stable after tested with Orthos).
After a certain period, I had to increase the VID to 0.8625 because I got random BSOD.
And now I have to increase again the VID to 0.8750.

 

It is an interesting observation, but I really doubt processors can change their physical properties so fast.

 

The default VID is 0.9875V, so 0.8750V is still quite low.

My friend with same laptop model and same CPU, cannot use my RMClock setting because my VID values are lower than him.

I think maybe the reason Intel set the factory VID quite high for allowance.

But I quite surpised with the increasement of the stable lowest VID after a certain period.

 

Just a guess, that minimal VID could depend from the environment temperature. You might have done your stability tests under other (lower) external temperature a year ago.

 

Well, just lowered my maximum core temperatures from 60 degrees Celsius to 50 degrees Celsius; I dropped my maximum multiplier voltage by 0.1625 and it's stable. Awesome thread.

Now my cool and quiet P8600 is even better. Not to mention battery life increases from lowering other multipliers.

 

Wow 60 celsius.... mine very rarely hits 45 unless it's under load, and that was before undervolting. Still, that is a nice drop if you were hitting 60 and now down to 50 - gratz!

 

This was my "maximum" core temperature after stress testing. I rarely got past 42 degrees celsius before I underclocked at my maximum multiplier setting. Now I'm at 35-38 degrees celsius.

 

RMClock and SIW have read 5C low on my last two notebooks compared w/ Everest, Core Temp and others I forgot. Still, < 60C is a day in the shade for a semiconductor.

 

Okay guys iam new to this so it will be a pleasure for you to help.
I testing my undervolts on performance on demand profile WHILE USING HIGH PERFORMANCE ON VISTA POWER THING. Now what happens if i change my vista power to power saver and leave the Rm clock on performance on demand?
What i dont get is does switching vista power to balance power saver have any effect on performance on demand. And how would i make the best battery life by leaving it on performance on demand and power saver.

Thanks guys

 

If you only changed the core voltages for your highest multiplier then you would only have undervolted for "high performance". When you change to "power saver" the computer downclocks your cpu, meaning you use less power (more battery), but also a lower multiplier. If you've undervolted your lower multipliers then the underclock should still be there--granted that you have RMClock running at the same time.

By underclocking the other multipliers you'll get the best battery life possible.

 

Ok got you thanks, lol my first bsod startled me, lol.

 

Just to be clear power saver locks it in at the lowest multiplyer for me which is 6x.
Whats balance lock it at?
ANd high performance?

 

Okay but how do i test 7x 8x

 

Put it at high performance and uncheck the highest multiplier box and hit apply in RMClock. Check your cpu info to make sure you're at 7x or 8x.

 

Why Speed Switching and Power Profiles Don't Save Anything - Chapter II

This time I found an old program from 1995 that calculates the value of Pi to a selected number of decimal points and records the time spent on the job. It's an old favorite of overclockers' called Super Pi. This is a pure CPU work (a little memory too) with no graphics involvement. I ran the program at two sppeds: 2095 and 1227 MHz (avg.) and recorded the completion times. The value of Pi was calculated to 32 million decimal points, a feat that required 80 hours and 38 minutes on a Pentium 90 some 14 years ago (God bless Intel !)

My stock fan, after undervolting, kept the CPU temps within 50 to 59C no problem.

At 2095 MHz it took 1423 seconds (23 min. 43 sec.)
At 1227 MHz it took 2259 seconds (37 min. 39 sec.)

The ratio of the job times is: 2259/1423 = 1.59
The ratio of the CPU speeds is: 2095/1227 = 1.71

These values are nearly a push and show, again, that working at half speed, a job takes twice as long and even using 60% of the energy (not quite true) shows NO NET SAVINGS.

Undervolt for overheating, yeah, but speed switching is for amusement.

 

I have a question about the undervolting instructions.
Okay so I am on the profile page to undervolt my computer. The instructions say to lower the voltage on the highest multiplayer index. For instance
my profile looks like this:

index Type FID VID
0 superLFM 8.0x 0.9000v
1 normal 6.0x 0.9000v
2 normal 7.0x 0.9625v
3 normal 8.0x 1.0125v
4 normal 9.0x 1.0750v
5 normal 10.0x 1.1375v
6 IDA 11.0x 1.2000v

for index 1-5 are checked(or boxed)
so my highest multiplier is 10.0x. So if I lowered the voltage on index 5 from 1.1375 to 1.1250 I test by running ORTHOS. Now the instructions say run it for 45 mins then if stable continue to lower it until the system crashes. Now do I only lower index 5? Or do I lower all the index a little at a time at the same time?

 

Tick automatically adjust intermediate multipliers and lower index 5 so you dont have to adjust the rest

 

I am using C2D T8100 2.1GHz.
I found that my EIST settings,
Enable Enhanced Intel SpeedStep Technology(EIST) was greyed out.
Is there any trick to fix that?
I hope some registry tricks could do that.

 

hey
if i put min processor state to 100%, what does it mean? my processor will be running @ 100% all e time? den isnt it going to be overworked/uses up alot of battery life?

 

oh and one more thingggg. this is serious. my dad's had his HP laptop for quite a long while. today i was gonna undervolt for him, but when i started orthos, to my amusement, e cpu speed was recorded at 767mhz instead of his 2.0ghz cpu speeed. den i ran RMclock, highest cpu speed was also 767mhz.. wad's happening?! why isit so underclocked? im sure he did nth to it cuz he noes nuts bout tech.. how can i make it back to default?

 

anyway.. last oneee! u wrote check only the last box, IDA. but isnt IDA supposed to be using only one core? shldnt i be checking the 2nd last box, which is the highest normal multiplier, 9.0x in this instance for me? and then also check the 2nd last drop down FID box if i want it to run at max FID all day long? cuz i only undervolted the 9.0x multiplier, i didnt test the last IDA undervolting, and its voltage is set at a default. plz advice, thx.

 

For people said that his stable VID increased over time. And someone answered that the higher ambient temp is, the higher stable VID will be. In this point, I think that the lower ambient temp is, the higher stable VID will be. Semiconductor material have conducts best at a certain temp. And that temp is a little higher than standard room temp (25C). Sorry if I was wrong.
And about 100% minimal and maximal cpu utilization in Vista, I think that mean nothing, because anyway, RMClock will take control of the cpu. I tried on my laptop, and cpu is still under control of RMClock. Hope this help.
Also, if want to see my undervolting result, you could visit my signature .

 

Very professional looking chart on your linked page. It says it all. Now, just a heads-up on some of the issues you raised. The heat conductivity of a silicon based semiconductor is flat or even or the same over a huge range. You could melt your notebook down before you'd see a change and maybe not even then. Also, VID's don't change with temperature. They are determined by the elements and alloys making up the junctions of the transistors.

The first thing that is affected by excessive heat is the insulating plastic film that separates the many layers of circuits that are coated onto the silicon wafer.

Cheers,

 

Go to advance tab, under intel core/core 2 family enchanced power states, deselect enable enh.halt(c1E). That should help.
I mean for the processor Ghz, you said it is lower than its normal rite? So, this might help you.

 

Its strange though my lowest multiplyier is running at the lowest voltage already, how can i get it lower.

 

Well, I am not sure. I would just expect more stability at lower temperatures, but I did not know there is some conductivity vs temperature critical point around room temperature that would change my expectation. Thanks for correcting me.

 

Well, as I understand it, that is the meaning of the min. state, but if RMClock is running it overrides this setting.

Now, I don't think the min. & max. processor states have anything to do with the VID spikes I had. Fact is that when I change from 'balanced' to "power saving" they are gone, even though my min. & max. states are different.
I'm noticing some other strange behaviour too btw, but I need more time to experiment with everything.

 

Jeez guys, you don't get to invent your own laws of chemistry and physics as you go. CPU's work just fine in liquid nitrogen at minus 300F right up to where their components start to degrade above 250F.There is no optimal anything. CPU chips get hot because they pull up to 30 amps of current thru a silicon wafer the size of a finger nail. Undervolting works because the stock voltage is set unnecessarily high at the factory and can be safely reduced a little bit. That's all.

Nifty explanation of CPU manufacture process:

http://www.cpushack.net/MakingWafers.html

 

Hello this is my first post here but I have been checking this site before regulary.
I want to share my undervolting experience though it isn't much as I got my first laptop a week ago and might do a short review or summary about it sometime if someone is intrested about it.
I used this guide and began with 10x @ 1.15v and got down to 10x @ 0.925v and have tested it with prime95 1h small FFTs torture test and I must say I'm very suprised about this as 0.925v is lowest value rmclock gives me. I can't say how much it impacted in cpu temp as it is stuck at one temp but mb sensors work ok and showed 5-15c drop orginal 50-60c and now 40-45c with prime 95.
I might try to install windows 7 soon and check if the cpu temp updating would work as it shows a different value (Between 20-40C) each time I restart or return from sleep but it stays there all the time after that.
I have tried rmclock, hw monitor, speedfan and core temp and all show exactly same temp for cpu.

 

When running test when using RMclock, under cpu info, near the bottom its says:

Multiplyer (FID) Current | start up| minimal | maximal.

For my current it says 6.0x however the example it shows 11.0x shouldn't mine say 10.0x since that's my highest multiplier? I feel like I am running the test wrong.

 

Why isn't this a sticky?

Great guide