The "Undervolting" Guide

so is there such a thing as an "accurate" temperature monitor for ppl to download and use ?

 

I think saw somewhere that some of the penryn temps are not accurate on RMclock (at least for mine). If theres ambiguity, pick a third monitor and see which one it agrees with. I use HWmonitor and it works pretty well, coretemp is also pretty good.

 

any ideas ?

 

thats odd, because it feels like RMClock is reporting correctly. when i check HWMonitor, it shows temps usually on the order of 7 C more than RMClock, and it also reflects temp changes too quickly (2+, 3+, etc.) for no apparent reason when my machine is idle.

for example, when i get the machine out of sleep, i immediately check RMClock and it gives a number which is very close to ambient (say 24/26 C). but HWMonitor tells me it's already @ 33/37 C.

i guess i'll download and try out everest, and see what numbers that tells me.

 

hey guys, i'm a n00b at undervolting so bear with me...

does this guide apply to AMD processors? i have the Turion X2 TL-52, so the Advanced CPU Settings screen obviously will look different from that of the Intel ones. when i check the factory voltages (i.e. by pressing default after unticking the box) for the various multipliers, i get that all of them are the SAME voltage, which happens to be the max voltage. is this normal? or am i missing something here? should i still carry out the undervolting?

thanks

 

Does it show the voltages and multipliers?

Its not accurate because it detects the Tjunction temps wrong for certain cpu models. Sometimes chips with a Tjunction of 105c gets detected as 100c, giving a 5c offset between programs.

I suggest using CoreTemp (no need to install). This will probably the most accurate program. You can then sync the temp offset using the RMclock registry.

Yes this guide will work fine for your processor. AMD just has little kinks but its worked very well in the past.

-Tick the 'Auto-Adjust Intermediate VID's' box
-Set the 4x voltage to .800v
-Set the 8x voltage to 1.050v

Start the stability test from these settings.

 

I get this:
I can change the nids but i dont have 7 or 8 indexes

 

Thank you for helping me

 

Those indexes reflect the underlying processor technology. Therefore, it is processor specific. In your case, there is only two multipliers. You may still proceed with undervolting.

 

Hey all, I got a few questions before I go headlong into undervolting again.

First off, what results can I expect from my T5550? Since it has picked up a rather nasty habit of overheating at stock voltages, I would love to start dropping to cooler temps.
Secondly, I just want to make sure that stress-testing with ORTHOS is enough. The last time I did any kind of undervolting, I was stable through ORTHOS, but had random system hangs when doing other far less CPU-intensive tasks. Any advice?

Thanks!

 

Hi you have a 2ghz processor which has a max 7.5x multiplier. Unfortunately, RMclock cannot recognise half-multipliers.

If you undervolt you will only get

FSB266mhz x 7 multiplier = 1.8ghz

Instead of 266mhz x 7.5 multiplier = 2ghz

No problem

You intermediate multiplier voltages were too low, thats why it was crashing on less-cpu intensive tasks. Less CPU intensive tasks use lower multipliers.

I cant predict what results undervolting will do to your T5550 since it varies chip to chip. Id estimate about 10-15c.

Have a look through the previous pages and youll find the info you need. (good luck!)

 

I'm getting the following error in Orthos after running the torture test for a couple of minutes.. What does this mean??
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

 

it means you set the voltage too low. try setting the voltage up a notch or two.

 

Hey all,

After following the setup for RMClock, I cannot seem to get my 6x multiplier to the lowest voltage possible, even though at the CPU Info tab, the possible voltage is .950V. Currently, my lowest voltage possible for 6x is 1.0625.

I have a T5550 if this helps any bit.

Can someone tell me why this is happening?

 

I guess im not sure if this is a problem with my undervolt or something i may not know about but here is what just happened to me.

I was stress testing my cpu after i had undervolted it a tad, i only nocked it from 1.375v to 1.000v. Now i was running orthos to see if it was stable i was going to run it all night (for 5 hours i have to be up for work soon) luckily i got up to get something to eat and peaked at my laptop because it was saying cpu overheating and then it kept showing the CPU throttling it kept bouncing from 50% to 120% i didnt take any screen shots i got scared at stopped it imediatley. I had only been running orthos for 10 minutues which i had done before without any overheating. but this time with the undervolt the CPU temps had shot up to 100C when before they would only be at 70ish.

I have no idea what happened but it scared the hell out of me if anyone has any idea what has just happened please let me know =D

Oh yeah my specs are C2D T8300 @ 2.4 GHz, 4gb ram, 8600gt, 250gb HDD. I think thats all you should need if you need anything else just let me know

And thanks in advance for any help/suggestions you may have

 

And I think I'm going to avoid undervolting my system.
When I tried to, TF2 crashed hard and corrupted my Vista install.

 

You might have missed a setting.

T8300 isnt supposed to go up to 1.375v. Max should be 1.250v

Are you sure its set/detected as a mobile processor and not a desktop cpu?

Can you explain abit more on what happened? What settings? etc.

In the Advanced CPU page, is your cpu set to Mobile?

 

My bad is was set at 1.1375. But this happened when i had it undervolted

 

i read more than 100 pages of this thread today and did some undervolting on my T7500 (HP 8710p)

the default voltage at 11x (2.2GHz) is 1.175 and the CPU temperature goes to 84.x - 85 degrees celsius with the fan running at maximum speed

i managed to lower that to 0.9625 and it ran stable for more than an hour with a temperature of around 72 - 73.x

dropping it to 0.950 slightly reduced the temperature - 71.x - 72.0, but it caused a BSOD after 25-30 minutes of orphos

my idle temperature with SLFM enabled is around 50

so, undervolting is great so far.. dropped my temperatures by 13 degrees, and the best thing is, the fan kicks in a lot less and never at full speed (looks like it needs to go above 80-82 to go there)

next thing to do is, observe the modes used to see if the intermediate modes (between idle and 11x) get utilized a lot and try to tweak them too.. i now have them progressing from 0.850 up to 0.9625, but i havent tested their stability yet

i'll try to see how it behaves on battery later too

 

Ohhh that's where I went wrong. I had it to mobile but apparently I didn't restart RMClock.

Quick question, does this affect the idle temperature much?

Thanks a lot mate!

 

yes, the idle temperature is affected (havent measured it), because of the fact that i enabled SLFM and set it to 6x.. so it's now idling at 600Mhz instead of 1.2Ghz

 

Amazing. I thought Fujitsu set the fan heat range so high (59-63C at idle) either to show the A6120 reviewers that they couldn't even hear the fan or to sell me a new laptop every year when the embedded CPU fried itself.

Sure enough, I can now run Orthos at top speed (2.0 GHz for T5750) and
hold the temp. to 67C using 1.000 volts! Wonderful. Buys me a year to find
another manufacturer who doesn't employ a short-order cook as a BIOS programmer. Thanks again!

Scott
a n

 

No risk for it to fry as it could go all the way to 90c without a problem. If the temperature goes any higher (120c I think) your C2D will turn itself off.

 

thanks to the wonderful guide.. i m using AMD TurionX2 TL-52.. my max temps went from 85C down to 69C.. WOW!! thanks guys.. but i m still experimenting since i got a BSOD just a few minutes ago while surfing.. hahah!!
my voltages are:
4x - 0.725V
5x - 0.7625V
6x - 0.825V
7x - 0.825V
8x - 0.85V

Didnt think that undervolting had such wonderful results.. thanks so very much!!

 

You're right. But in the long term, months and months of high temperature
operation and large temp. swings (delta-T) causes metal fatigue and eventual cracking in integrated circuitry using nanometer-thin wiring. I use the 2 C's rule: "Cool and Constant" in electronics. My car says 6000 rpm
is redline, but for how long? Same thing for Intel chips. I think notebook manufacturers are sacrificing durability for low fan noise. Cripes, I can't even hear mine. And, on my old Uniwill with a Northwood P4 pulling 100 watts, I was happy to hear it run (at 45C)! Like BIC lighters, we're on our way to disposable laptops.

Thanks for the feedback.

 

I started off in this undervolting biz a few days ago like you. What all of Fujitsu tech services couldn't be bothered with, I fixed in 15 minutes with RMClock and the "undervolting guide". I used it like cruise-control in my car, start off on low speed (1.0 GHz for my T5750) and at 15% load, it kicks in at 2.0 GHz (at the same voltage 0.950 hee hee) until the workload is over and then it drops back. I don't even bother the CPU with intermediate steps.I don't need a 6 speed transmission........ Wish I had SLFM too.

Best of luck.

Scott

 

Heres what i got so far but it wont let me lower the voltages any more..

but i did go from 61C to 60C... but the V went form 1.35 to well the one showing on cpu-z but it seems like the software isent chnaging it all..

 

When you go to the "monitoring" tab in RMclock what voltage is it running at? I would trust that number.

 

ahh crap when i try to make a new profile i need to buy it... so when i restarted my laptop and went back on it all went back to default...

ok re did the seetinsg and this is what it shows.

nvm i ermm got it for free....

 

You missed this part of the guide:

Any version after v1.41 will show the desktop equivalent voltages of your CPU. It seems they have stopped support for newer notebook cpu's (merom/penryn).

For older cpu's (yonah's) any version of CPU-z will show the voltages perfectly.

 

Something weird. Orthos running on my brand new T61 just keep saying Paging, and even after 25min I doesn't do anything.



<a href="http://img604.imageshack.us/content.php?page=blogpost&files=img527/3478/orthospagingrx8.jpg"

Is this a fault from the HDD, I just got this machine new, everything is functional, is just that when stressing the CPU which is a T8100 the HDD LED indicator is ON all the time also the temp doesn't shoot up.

IS this normal?

 

i can see that my cpu isnt always running at the specified voltage for that frequency..
i've got 11x (2.2Ghz) set to 0.9625V, but it sometimes spikes up to its default voltage - 1.175
strange..

 

well mine is saying paging but its defo using all the CPU as task manager and Rmclock monitoering says so and your right CPU-Z is displying it correctly now thanks you.

 

I created the Recovery Discs for my T61 using R&R, and the proceeded to completely wipe off Vista and the recovery partition. Reinstalled with XP Pro.

What a difference, having only 1Gig of RAM makes Vista monstrously slow (I have never liked it anyway).

Using Orthos in XP works fine, I'm now testing how much high the CPU temp spikes up, and seems I 'do not need' to undervolt. 10 mins has passed and the CPU is only 56C with stock voltage (***) this is my coolest Thinkpad ever.

 

u might as well undervolt to help battery life. i made my 44% batt life last 40 mins lol trying to see how fast i can make it go down

 

I have a Intel Duo Core T2300e (yonah) @ 1.66GHz
Now i know that the cpu consume power in watts. If I reduce the voltage, the ampere will increase. Could the increase of the ampere damage the motherboard or the cpu?

 

I find now an issue interpreting temperatures.
Troubadix, the author of the awesome program of TPFanControl for Thinkpads, report that the THM0 THM1 Core#0 Core #1 are the same values for the CPU.

I followed the guide to undervolt my T8100 45nm processor here at NBR. Then installed TPFanControl v0.50. HWMonitor temperatures from the Cores 0 and 1 are different than the 'CPU' reading on TPFanControl.
TPFancontrol reads the THM0 value as CPU. In my opinion right now in the T61 is useless to control the fan because we're getting a totally different reading.
Here, take a look at this screenshot.


You can clearly see there's discrepancies on the CPU reading.

Yesterday I wiped the HDD that was preinstalled with Vista Home Premium and downgraded to XP Pro. I don't know now why my fan won't stop working once it read the threshold to start spinning.
I remember in Vista the fan would stop but in XP is not.

I've been working with your 0.18 version (quite old) on my T42p for quite sometime, and now I'm unfamiliar with the new values the TPFanControl has:

1 cpu 42°C (0x78)
2 aps 41°C (0x79)
3 crd 34°C (0x7a)
5 no5 32°C (0x7c)
7 bat 30°C (0x7e)
9 bus 38°C (0xc0)
10 pci 40°C (0xc1)
11 pwr 37°C (0xc2)

against
ACPI
THM0 42C
THM1 30C
Intel Mobile Core 2 Duo T8100
core #0 31C
core #1 28C

What's the CPU temp where the fan on the T61 should stop?? I'm using an 4:3 SXGA+ model.

-- Later troubadix sent me his image with the temps readings on his T61:



Which value should I care about if I'm seeking the max CPU temp when stressing out?

 

No Sir !
Ohm's Law states that E=I*R where E is volts and I is current (amperes) and R is, of course, resistance (ohms). Furthermore, Watts=E*I.
So, lowering the voltage across a fixed resistance (your CPU) lowers both the current fllow AND total watts of power consumed. That's good because for a simple resistor 100% of wattage is converted to heat (the enemy). Motors are different because most of the watts consumed are converted into work and only about 10% into heat. Of couse below a certain voltage transistors quit functioning and you get the classic BSOD.

Woohoo!, I can still remember my high school physics after 40 years. Theres still a few neurons firing in the old noggin...........

 

so this guide is based on a santa rosa platform - i assume it works in exactly the same way with a montevina? i'm thinking of undervolting my sager 8660 when it arrives

 

So I ran the Orthos to learn the max temp of my CPU, and it maxed at 58 degrees. Should I still attempt an undervolt?

 

As scott.ager mentioned (i hate getting technical). Undervolting has proven to be safe over the years.

Have you tried using CoreTemp? (no need to install)

I can see a 15c offset with RMclock > TPfancontrol and 5c offset with RMclock > Hwmonitor. This happens when the Tjunction max is detected incorrectly from the MSR (model specific register)

Yes it will work for the montevina chips but there is a small catch for the P8400 and P9300 cpus. These cpu's have a half-multiplier which RMclock cannot detect meaning it will slightly downclock if you undervolt.

Entirely up to you, if your happy with your notebook then why bother.

You wont see much of a temperature improvement since your notebook is already running cool to begin with. (must have a good cooling system)

You will see a bigger gain on battery life if you do undervolt.

 

Weird, I undervolted my T8100 processor on T61:
I just put the lowest value in all Multipliers and the machine is stable. Is this OK, it seems absurd to have a 11x Multiplier run at the same voltage as the 6x:

 

Yes thats normal to happen. If you have a look in the previous pages, youll notice that a few users were able to use the lowest voltage just fine.

T8300 and 8100 seem to have the best voltage tolerances.

 

I might have to give that a go at some point with my T8300...

 

thanks for the warning - luckily i have this: Intel® P9500 45nm "Montevina" C

i'm getting the same laptop as kevin - it should arrive today. i think i'll be undervolting for that very reason

 

im not sure i understand the risk(s?) if undervolting. wasnt it dangerous for you to set the voltage to the lowest value? doesnt that mean you might get the BSoD (blue screen..)? and the BSoD is permanent, no?
thanks for answering all these questions, i'd just like to be sure before i risk such a big investment!

 

Woops i meant P9500 not P9300 (it doesnt even exist). I confused it with a T9300

P9500 has a FSB of 266mhz and a 9.5x multiplier

266mhz x 9.5 = 2.5ghz

Since RMclock cannot detect half multipliers, it will only see it as a 9x multiplier

266 x 9 = 2.4ghz

You will loose about 100mhz if you undervolt.

No the BSOD isnt permanent. The BSOD you get is a system instability warning which will create a memory dump then restart and restore the previous settings. The undervolting is done through software, which means its not permanent. It will only load when windows load.

 

Thanks flipfire. Weird how on my T42p the minimum voltage it could get is 0.716V, but at the same time the laptop was much more hot than this T61.

I'm glad I got the T8100 most T61 come with old T7xxx series with 65nm technology.

 

How long did you do the stability test for at 0.950v?

I suggest raising your 10x to 0.975v. This will ensure uncompromised stability.

 

aw.. well would i be able to set a shortcut to undervolt for when i need the extra battery life and to restore it to normal when i need the extra processing power? I dont know if constantly switching this would be healthy for the computer, but if it isnt bad, i would love the extra battery life