The "Undervolting" Guide

I'd consider myself a "noob" concerning this type of stuff, and with the exception of my laptop being finicky in it's old age, it's been pretty easy for me to pick up on. I've had a couple issues here and there, but nothing detrimental and I think it should be easy enough to follow along in the guide for anyone that has moderate skills. Since you have an Intel, I'd expect you to have an easier time of it than me. I don't think I recall anyone else on this thread using AMD aside from me, and seems my questions were brand specific.

 

If undervolted too low, does it give any specific BSOD?
I've been receiving these BSODs (probably from hardwre) lately that I haven't figured out yet ...

 

I don't think there are specific BSODs caused by indervolting. If I'm undervolting then I would consider that to be the likely suspect. All you need is for one electron to misbehave and then the system can get confused. It is prudent to give yourself a voltage margin of at least two steps above the minimum voltage which survived a long stress test.

John

 

I should draw this to the attention of any 64 bit RMClock users (link to signed RTcore64.sys).

John

 

Thanks John. Good stuff, works a treat.

NEWS UPDATE for x64 undervolters:

We have finally got hold of signed drivers. We can now scrap astiv driver bypass and use RMclock normally!

Please see guide again for the download link.

 

Im using the PHC Kernel-Patch and the Gnome Utlity to Underclock. Google should help. The numbers are standing for a Core-Speed. On my old Notebook there were numbers from 6 ascending up to 16. Makes more sense than that high numbers. but it does not matter because it works! I guessed that the first number stands for the highest core speed. But i can not exatly tell how to calculate the number. You can calculate the VID (the number after the colon) for Core (2) Duo chips: 712.5 mV + VID * 12.5 mV = X (Voltage you want to set).
That means: (X - 712,5mV) / 12,5mV = VID
Example: I want to set my Voltage to 925mVolt (0,925V)
925mV-712,5mV = 212,5mV / 12,5mV = 17

So my VID for 925mV is 17. I have to write it after the colon for the corresponding core-speed (what is coded with this other numbers...)

 

mhm, not sure if it's already answered in this huge topic, so I'll just ask it again.

My xps1530 (t7500) has all the multipliers on 1,2500v... Can I just lower them all at once? or what should I do?

http://img139.imageshack.us/my.php?image=rightmarkxps1530cz0.jpg


Thx in advance


I really hope it will help with my heath problems... My laptop goes to 90+C degree when stressed... (during not really very demanding games like for instance World of Warcraft)

 

Sweet just downloaded the signed 64-bit driver, thanks again flipfire!

 

Your voltages were detected incorrectly. Did you hit the default button?

If not just use the default voltages from the guide (its also a T7500)

Yes it will definately help with your temp problems, hopefully by more than 10c

 

Thx, I used your voltages, and the temps already dropped by 10 degree.

Then I found some t7500 voltages on the internet, I tried them, with the voltage for the 11 multiplier at 1,0375. But that gave an error during Orthos testing. (no BSOD though, just ORTHOS)

Using CPU #1
Beginning a continuous self-test to check your computer.
Press Stop to end this test.
Test 1, 4000 Lucas-Lehmer iterations of M18474367 using 1024K FFT length.
Test 2, 4000 Lucas-Lehmer iterations of M18274367 using 1024K FFT length.
Test 3, 4000 Lucas-Lehmer iterations of M18274369 using 1024K FFT length.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 6 minutes 33 seconds - 1 errors, 0 warnings.
Execution halted.

Now trying with 1,0500.

Temps went already down from 92 to 81. It's a start

 

I'm going to give this a try with my XPS m1530.
I had to to this to save this in my UserCP area.

 

I noticed when I do CPU stress test only in Orthos my T8300 can handle 1.0250v. But when I do blend it BSOD's right away. You think my ram is at fault?

 

I ran Orthos 10min and it only gave 69ºC of max temperature.

Should I undervolt? Or is it ok? (T9300, M1530)

 

Yeah why not? I had 65ºC max and undervolted, now I have 45-50ªC. I'd say the less heat the much long processor life..

 

Check if rmclock is set to "mobile cpu" (advanced settings).

ive also got a xps m1530, but with a T9300. what temps doues your GPU reach (idle and whie playing)?

 

Well I tried undervolting (T9300, M1530) but it gave me tha same max temperature (69ºC), I don't know if I did it correctly though...


(Normal VID for 12X was 1.375V)

PS: I am also Spanish

 

Did you followed the guide step-by-step? In profiles(your screenshot) you should change Start Up to Performance on Demand. If not on system restart RMclock starts up but DO NOT load Performance on Demand profile. Checkout the guide again.

I'm not, I just live in Spain. Pero encantado de conocerte

 

Yes, I think I read it completely. Anyway, I'll try it later or tomorrow (I don't want to stress my new machine too much)

Thanks. Y, lo mismo digo

 

well, with my xps 1530 T9300 (same config as JinRoh), i get temps like 42-44 C° when Idle (with Ubuntu). In Windows the temps are a bit lower. I think the main problem is the increasing GPU temperature because the 8600 is not completely supported yet in ubuntu (my guess). I can see how the GPU-Temp increases up to 65 while working in office, then the CPU increases the tep up to 51° and then the fan starts working.

But: Try to change the (i do not know how you call it in english) thermal-paste (?). Dell uses pads that are not so effective. Im usind ArcticSilver 5.

 

Hmm which voltage is correct?
CPUz : 1.113V
RMclock : 1.000V

Maybe CPUz still dont fully recognize that I have a mobile chip? Its the latest version 1.45.

 

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3184&p=5

I'm not sure if this answer your question but it'll make no harm if you check it out..
My point is that RMclock shows you the actual voltage, and CPU VID is..something else

 

My main problem was a clogged fan it seems. I cleaned it out, and my temps dropped by almost 20 degrees

Now, with undervolting, I reach max 70degrees fully stressed.

Still rather high imo, but I guess for a notebook, that's as good as it gets.

 

ASUS S96S T7300 2.0GHz
Default - 1.175V
Undervolted to !0.925V!- rock solid!
before- 82C full load stress test, now 61C(60.7) max,
preeeeeeeety good!!!

 

I have no idea why, but it is not allowing me to click Save File when i try to download rmclock

 

try this link http://www.mediafire.com/?1i19cvcy1zl
Check firewall/adware/spyware/popup blocker/firefox blocking extensions etc

 

is there like a mid point between under volting and overclocking together, or should you just avoid that completely.

or is overclocking, then undervolting and testing for stability completely safe?

also does undervolting carry on if you lets say, replaced your harddrive

 

I can't say I understand this question at 100% but yeah you overclock first then undervolt and test.

No. RMclock(or whatever other program you are using for undervolt) must running with profile loaded. If you replace the hdd, and you don't install & configure rmclock again, you will have the default factory-made voltages.

 

for some reason, my initial testing lost me 15 minutes of battery life.

T9300
1.1375 is what it starts off at, 1:45 hr battery life. x12
i got it down to...
1.000 right now,(haven't tested stability fully yet, but it bsod'd at 0.9875 and i only get 1:30 battery life...

heat did go down though, but what about battery! :/

 

It is possible that RMClock is conflicting with some other power management. Look on RMClocks battery page to see the power drain under different conditions. Are you using the Performance on Demand or another profile?

John

 

Is it bad that my 12x multiplier is 1.0750 then all the others are 1.3750? But SuperLFM is .9500... It's not checked either.

ok here is what it looks like after running orthos:

 

I'll recommend you to read the guide again, all of it . For start if all multipliers are at 1.3750V then the voltages aren't detected properly. Hit the default button.
If you undervolt just 12x multiplier, it meens it will use less voltage(less heat etc) only when your cpu works at 100%.
SuperLFM-->Check the guide. In two words - idle voltage.

 

Ahhhh all right. I didn't notice the default button instruction part. Thanks!

 

Also, if your 12x, or whatever your highest multiplier is, works at a certain voltage, it's safe to set any of the other multipliers that were higher to that voltage.

 

I am trying to undervolt an AMD Athlon 64 x2 Dual Core Processor TK-53 1.70 GHz.

Right Mark reports it as an AMD K8 Family advanced CPU.

None of the settings look the same in the "Advanced CPU Settings" page as Flips pictures in post #1 of this guide. So I leave everything as is on that page and tick apply these settings at start up.

Go to Profiles / Performance on demand page and tick all right boxes, go to Profiles and chane current to Performance on Demand for AC / Battery, un-tick Auto Adjust ....

... AND when I hit apply I get a flow of flickering horizontal lines from top to bottom on the right 3 inches of the screen. This is before any tweaks. Am I missing a set up proceedure that is different for the AMD processors.

 

maybe.. i get the same/ if not less battery life when i undervolt, and the idle temp is 1-2 degress lower only, max temp is like.. 4 degreses lower :/

 

Back to working on my XPS m1530.

Does it stand to reason that the lowest usable voltage (with no BSOD) on the highest multiplier, could also be the lowest voltage for all the other multipiers.

Right now my 12x (highest) is working at .9875, but that is lower than the next 4 setttings above it. It is not until 7x that it drops lower. Could I run 8x-12x at .9875 safely .... OR Does each have to tested seperately?

 

Yes, you can use the other ones at that safely. If your highest multiplier, which is using the highest power draw, runs at that voltage, anything requiring less power will also run at that. My T7500 (2.2 GHz) runs at .9875 also.

 

OK I see that there are a few Undervolt Penryn threads, but this general undervolt thread seems to be still the most active.

I am noticing a few odd things in Right Mark

1. On the CPU Info page it is not reporting my T8300 as a Penryn. CPU Core and signature are reported as unknown. As far as I know, I followed the Guide on page 1, but maybe there is some extra setting for penryns.

2. When I tick only SuperLFM and set it for 6x I get 3 different Mhz reading from 3 different programs: (not running Orthos, it is just sitting idle for the reading):

- RightMark says each core clock is running at .... 598.5 Mhz
- CPU-Z has the core speed at ........................ 1197.1 Mhz
- Orthos Beta has CPU speed at ....................... 2393 Mhz

Who has it right here? Or all they all 3 right from different points of view?
-RightMark for each core
-CPU-Z for the total of the 2 together
- Orthos for the Total capability of the T8300 (not seeing what RM is doing to underclock the processor)

 

Tis odd, but it may be due to the 8 series cpu, could have id issues. My 7 series all match up within .1v and 300mhz in assorted monitoring programs. I believe RightMark gives you what the levels should be in a more theoretical since, as It says my voltages are right on the money, but I can't believe that in actuality they are 100% perfect, CPUZ agrees with me

 

Yeah no problem, but you should test every step. The point is to achive better results. As for the Penryns I'd say RMClock does have issues with it, I've seen some posts from user complaining.

What do you mean? CPUZ shows you the actual voltage? Because it doesn't.

 

Try CPU-Z version 1.40.5. It is older, but seems to be better.

John

 

For some reason, my other laptop`s, T2300 doesn`t want to UV. I`ll have to look into it.
Is there an issue with Yonahs and UVing that I don`t know of?

 

What do you meen "doesn't want to" As far as I know RMclock do ok with yonah...

 

I've had not problems in the past undervolting a T2250 and T2300.

Make sure you have selected mobile in the CPU type and then restarted RMClock.

John

 

I already know the story.
It`s just not remembering the voltages. I set them somewhat lower,ran Orthos, made sure the "performance on demand" is selected for both battery and AC mode and still it keeps the default voltages.
It`s an HP Nx9420 with a T2300 1.66 Core Duo. I also had issues with a 2.0 Core Duo.
Hmm, I`ll need to double check everything again.

 

Did you actually hit the Apply button after setting the new voltages?

I did some more testing yesterday and dropped mine down to 1.175 and tested stable for about 20 min in Orthos (all the time I had to spare). My temps didn't seem to drop much (actually ticked a degree higher for a split second) from the previous votage. But it was a hot room I was in, not my usual cooler area, so I doubt the temps would be acurate for testing. It was in the low 40's at least.

I also noticed something very nice about it that I didn't expect. I used to not be able to play games in my truck with it shut off or my power inverter alarm would go nuts as soon as I started the game on the default voltages. Last night I played a pretty demanding one for my system for about an hour or so with my screen about 50% before the low voltage alarm even beeped once. I've got 4 deep cycle batteries, so I can go for hours with light use without draining them now. I knew I'd see a lesser power draw, but had no idea it would be that much.

 

There is a potential significant reduction of power drain (I have measured about 10W) when the CPU is under full load, but this tapers off as the CPU load reduces and will be insignificant under light load (no drop in voltage at minimum multiplier).

John

 

What the..I read eleron911 post and check out my config to see if I discover something. I open my RMclock(all set up and everything) and all my config is gone! I had all multipliers set to 0.950 and rockstable system, now I have it like starting for the first time, no profile selected no profile configured!
And, like Mortal Combat FATALITY, RMclock measure my cpu temp as 23ºC ?! I've never ever had any temps below 40ºC so obviously somethings wrong..Ok, diggin into the problem now..

Edit: HWmonitor actually agrees with RMclock..wait..I upgraded my bios 2 days ago you think this can be the reason? I'd say probably, for the lower temps maybe - well 20ºC lower..sound like crazy, but this can't erase my RMclock config can it?

Edit2: Running Orthos for 15 minutes I actually reached the 45-47ºC which I had in idle(6x 0.950v).

 

Well, I got it working somehow,dunno what was wrong there.
Atthegates, if your voltages are low enough, >50C is attainable.

 

Yeah well..I was kinda freak out..