The "Undervolting" Guide

proof enough? edit: i figured it would be obvious as i dont have the stock processor in this, but im using AS5 as well

 

NO
When I had that BIOS issue, I tried several programs(CoreTemp, Speedfan etc.) and all were reporting the same - 10-12ºC. And I'm talking about T2370 dualcore, not core2duo or penryn. So excuse me if I'm too sceptical

Don't get mad I hope your temps are reported correctly, if so you achieved awesome results, maybe the best penryn undervolting so far..

 

A few more inconsistencies to be aware of, at least with T9300's.

For me, at any rate:

1) CPU-Z 1.45 reports a minimum voltage of 1.113v no matter what I set voltages to in RMClock. It accurately shows higher voltages however.
2) RMClock shows my voltage at no lower than 1.000v, regardless of what I've set VID's to. It reports a minimum voltage of .9250v.

I can demonstrate CPU-Z's voltage reporting to be inaccurate - at least in the 1.113v - 1.000v range: Under full load at 12.5x multiplier, I'm stable at 1.1000v(RMClock reported voltage; 1.113v CPU-Z reported). If I reduce that to, say, 1.0500v in RMClock, Orthos freaks out instantly and I BSOD in moments. Thus, clearly, the processor is being undervolted beyond what CPU-Z is reporting.

I'm curious, then: is RMClock is reporting correct voltages? Am I running below 1.0000v on SLFM when it's set to 0.9500v, or am I running at 1.0000v? I understand various processors/chipsets have fixed caps for undervolting.

Time to download some more voltage monitoring software to compare results.

Edit: Righto. Everest reports the same numbers RMClock does, so I think it's just CPU-Z that has no clue as to what my processor is actually running at voltage wise. Do any other T9300 users have this issue to, or am I just *special*?

 

what temperature is ur room?

 

Yeah, it surprising that his temps are so low, room temperature is like 21-24 C and you would think it wouldn't go below that. My T8300 hits a minimum of maybe 27C after I take it out of sleep mode

 

i would say that the temperature in the room is around 60-65 farenheit. mother in law likes the house cold....

 

Hi,

I have a T8100 on Toshiba Satellite U405,

I cannot undervolt lower than 9.5 volts in rmclock, and in NHC I cannot change voltages,

Intels official documentation says lower voltage range is below 9.5, unfortunately rmclock cannot detect my cpu an says "Unknown Core" like rtrdogs.

Hey rtrdogs, how did you managed to undervold below 9.5 volts ?

Cheers,
ilker.

 

diablo85, that's an amazing idle temp you got there.

 

Out to anyone with stuttering issues when using the 174 series laptopvideo2go.com drivers and RMClock. Anyone have a suggestion to start at for trying some new drivers. I just reinstalled Vista fresh, so I was thinking about going to 169.04, but its horribly hot compared to the 174.31's i was using, until they displayed a horrible lockup and stuttering fest. Anyone with an idear to start with? Thank you.

 

well thank you

 

Thanks for the guide (+rep). Undervolting has made a huge difference.

What kind of results are people getting with their AMD Turion X2?

My Results for the 1.8ghz TL-56 are:
1.8ghz (9x multiplier) 1.075v to 0.925v
1.6ghz (8x multiplier) 1.050v to 0.875v
1.2ghz (6x multiplier) @ 0.800v
1.0ghz (5x multiplier) @ 0.775v
0.8ghz (4x multiplier) 0.800v to 0.700v

I'm very pleased with these results, the difference in temperature is huge (at least 5 degrees and less fan activity) and it seems just as stable as normal. Do you think there is much more room to undervolt?

I've overclocked to 2ghz @ 0.925v and I'm testing it at the moment. I done a 1024m test with wPrime, then 40 minutes of Crysis with the GPU overclocked aswell (60 max CPU, 51 degrees max GPU). Now I'm typing this with absoutely no problems. The fan is barely active, and I was a bit worried that because the fan activity is linked to the CPU temperature, that whilst gaming it wouldn't be as active and the GPU would run hotter. The fan wasn't running as fast, but still it has only increased the max temperature by 1 degree (normally 50 degrees max).

So far it has froze 2 times and cut out (instantly switched off) three times aswell, but none of these were whilst stressing the CPU and the instant switching off has only happened from idle. I think it is because the voltage was too low. Which is strange because the first time it cut out was at 1.0v with 1.8ghz, and I have now gone to 0.925v and I haven't had any problems so far. I tried 0.9v but it would only idle 10 minutes before cutting out; it seems idle is more of a test than the stress tests. 2Ghz overclock at 0.925v has been fine idling for the last half hour so I'm confindent that's ok.

Is there a way to allow RMClock to startup with Vista and not get a UAC notification from it?

 

1. You're not underclocking, you're undervolting. The down clocking is part of the CPU, you're not forcing it to run lower than it's minimum which would happen in underclocking.

2. Who knows how much more room you have to undervolt, the only way is to stress test. And the worst that can happen is a BSOD, though you should only have 1 or 2 at worst since you'd bump up the voltage once you get one, right?

3. I wouldn't worry about the heat since the GPU heat will pass a bit along to the CPU, plus CPU's aren't meant to be at as high of temperatures as GPU's (ie. GPU's can get hotter than CPU's can before going into the danger zone.)

4. Did you remember to uncheck all the options? Because the same thing happened to me, it was always when I came down from load that I rarely had a BSOD, however since bumping up the voltage I haven't had another BSOD. Don't know if this is in the guide or not, but check with Orthos (small FFT's I think to stress CPU to the max) instead of any other program since it was the only one for me that caught the occasional error causing BSOD. Also, make sure you let it run overnight since errors might not show up for a while.

5. I just turned of UAC since I know what I'm doing.

 

1. Typo. I know what I'm doing.
2. I thought some people with AMD Turion X2s could say their results to give me an idea what's possible so I can get a rough idea how much more I may be able to undervolt (I've been searching the forums aswell). But I think I have found my min voltages now, definately for 9x multiplier atleast.
3. Ok.
4. I prefer not to run/install any programs unless I feel I have to. I think wPrime and testing with games should be ok. What options need to be unchecked? I didn't see any in the guide.
5. I'd rather keep UAC on for security. Isn't there a way you can make a rule to always allow it?

 

Other TL-56 users got around .900v which means your right on target. If you remove your OC then you can probably use .900v

Your idle is too low! If your gonna lower your idle, then you need to stability test that too

 

I was idling at the same speed to test them individually, so it was idling at 1.8ghz at 0.9v (no OC) and it cut out twice after 5-10 minutes idle. 0.925v is fine with or without the overclock, it hasn't cut out and has been totally fine.

 

This is too low (this is your idle). Causing this:

 

It may be; I've only stress tested 4x @ 0.7v, I'll test it at idle later. But I only had the 8x multiplier selected using "Maximal Performance" so it wasn't going down to 4x. I was watching with CPU-Z aswell, so it definately wasn't that, it appears to have been that 0.9v was too low at 8x as it doesn't do it at 8x with 0.925v.

 

Ok heres a resource for another TL-56 UV'd http://forum.notebookreview.com/showthread.php?t=209805

I suggest you bump your voltage up to .950v just to make sure it is fully stable. Raising it by .025v will barely make a difference on temps or battery life, you will gain more stability in the upside.

 

I found that thread earlier, thanks. He's got pretty much an identical laptop to me and appears to get the same sort of results.

Yeah, it is probably a good idea to add 0.025v, but the current voltages seem completely fine so I'll see how much temperature it adds first. Thanks for your help.

 

Is it 100% necessary to test the maximum temperature? I'm just afraid as I'm using a new laptop. But I want to undervolt it to preserve it's battery life.

 

Well yeah..you want something to compare results...and you must stress the CPU anyway to test the voltages..

 

Hi, i just got a brand new sager 2092, and when i ran RM clock it says that my CPU is running at 1250Mhz, when i have a 2.5Ghz T9300 CPU. It also says that im running at 6x multiplier when i should have 12x. Any clue whats up?

Heres a SS of my RClock: http://img167.imageshack.us/img167/7622/14844386fj0.jpg

 

I really really wished the search function worked.

NCG it is fine, that is what is called speedstep. Follow the guide.

 

It has been awhile since I posted in this forum, I usually post more in the Tablet Forums... I saw your guide and gave it a shot. I must say I am incredibly happy with the results....

Fujitsu T4220, Intel Core 2 Duo T7100 1.80GHz, 3GB PC2-5300 G. Skill, 120GB Hitachi 5400.3



At top speed the Voltages fell 12.77%, Average Temps fell 8.39%, and Average Battery Discharge Rate fell 19.86%!!!

Thank you so much for this great guide! I really hope my new HP tx2500z can pull of similar figures when it arrives. I'll post details again once I have completed the process.

Wow...

I got the temps from RMClock's logging feature and took the average of a one (1) minute sample after five (5) minutes of running OCCT with RMClock's built-in polling intervals.

The battery life was pulled from a trial copy of Passmark's BatteryMon. It was the average of a (1) minute sample after five (5) minutes of running OCCT with one (1) second polling intervals.

 

Thanks for the documented benchmarks. Never really had a chance to benchmark the discharge rate. At least now i can put an estimated figure on it, 20%.

Im afraid you wont be able to undervolt your new tx2500, due to the puma platform. RMclock most likely wont recognise it at all. Rightmark has been way behind schedule with their program updates.

 

Hello everyone!

I'm currently undervolting my new AW M15x with the T9300 2.5ghz Core 2
So far i've tested the top multiplier down to it's lowest stable setting (1.0125 down from 1.1375) and had a nice 10C (at least) temp drop at load.
Now i'm working on my next lowest setting, and i've just finished testing 0.9875v and it's stable.
The whole point of this post is this:
There are only 4 more voltage drops I can do through RM clock (stops at 0.9500v). Should I stop at any vertain point? And if not, is it fine to have a few multipliers all at that lowest setting?
Thanks!

 

I don't think so. As long as it is stable, you should have no problems if the bottom few voltages are all equal.

 

WOW

Great tut.

And thanks A loooooooooot

My temps 91 ----> 68 C !!!!

I cant believe it my self!!

And that's only 0.1 volts drop

Gonna try another 0.1 v.....lets see what will happen

Does this affects the GPU temps (I now it was said in the tut but i want to make sure)

 

Yes you can go as low as you possibly can. But to ensure stability i suggest leaving a voltage margin of 1-2steps higher.

If you can run ALL multipliers on .950v, then you are the very lucky.

If your gpu and cpu share the same cooling pipe/heatsink then Yes.

It will make your whole notebook somewhat cooler

 

i'm new to the forums since i also just got my laptop last week. i have sucessfully undervolted an acer 6920g (t5550) to its lowest possible voltage of .9500v for the max multiplier. max temp went down from 64C to 37C under load!

 

Good for you..if you have the max multiplier set with a lowest possible voltage this means you can set all multipliers to 0.9500..at least it worked in my case.

 

if anyone is running Vista with UAC on after this undervolting, i know how annoying that UAC prompt is at startup every time the computer starts! i was once reading in some other forum on ways to skip this and found a utility that you can use to make RMclock run as admin at startup without being prompted. just make sure that you select startup registry as option in the RMclock before using utility, then after using utility, simply go back to RMclock and remove the startup option so you wont have 2 RMCLocks starting!

http://www.jimmah.com/vista/Administration/autostart_admin_program.aspx

 

Does undervolting my notebook slows down the speed of applications (like loading time, etc)?

 

No, it says it somewhere in the first page.

 

I have recently undervolted my Lenovo T61p with a t8300 penryn. I got similar results as others have previouslyt in the forum:
12X=1.025V (original was 1.1375)
11X= .975V
All others = .95V

My lappy was already pretty cool before the undervolt (37-40 idle, 63-64 after 10' of full load with orthos). After the undervolt the idle didn't really change, and I got about 3-4 degrees cooler when under load. It doesn't seem like a big deal but when running the stress test the temperature rises noticeably slower when undervolted.

However I have run into one issue with RMclock which seemed to happen randomly the other day when I was switching Rmclock profiles and have not been able to fix it. I tried uninstalling and wiping the registry of rmclock, and reinstalling the program and it has not fixed this problem.

Now when I undervolt, the voltage settings do not always take effect and it seems to depend on what power management plan i have selected in my thinkvantage power management program (which leads me to believe it is interfering in some way with Rmclock). For instance, when using "energy star" in thinkvantage which allows for speedstep to be active, the cpu is reporting to use the default higher voltages for each multiplier. It uses these default voltages during normal computer use, EXCEPT when i use orthos to stress it, and then it reverts to using the lower voltage i set in RMclock

Whats interesting is if I use lenovo's "maximum performance" plan which runs the processor at the highest multiplier all the time, the cpu uses the lower voltages I set in RMclock and the cpu alternates between the different multipliers (i.e. it uses speedstep as defined by rmclock, and the cpu is NOT locked into the highest multiplier).

Sorry if this is confusing but I find it strange that rmclock might be superseded by my computer's power management for one plan and not the other. I'd rather use the energy star plan because it has other power settings (like screen brightness, sleep timers etc.) that I'd rather keep. Any suggestions?

 

Skip the thinkvantage program.
You can adjust the screen brightness manually, and it`s best to make the CPU work as YOU want it and lose some conveniences,than using those conveniences but using the CPU without control.

 

what's the lowest voltages people have gotten on a t7500? currently, my t61p is running stable at .9875V on the 11x multiplier...

i'll have to test out the thinkvantage program. I think I just have it set to balanced....not sure if that will interfere with the voltages...

 

Hey guys! I have an HP Pavilion dv2815nr (AMD TL-60 2.0GHz) , and when I go to the Main Profile page, all my multipliers have the same voltages: 1.1750V.

The guide says that undervolting all the multipliers are optional, except the lowest multiplier, which you should not change. However, what about gaming and surfing the web? Are you saying that gaming will be undervolted and surfing the web will use the default 1.1750V??

But yeah, I was just curious why RM Clock is displaying the same voltages for ALL my multipliers. Thanks!

 

Nice, mine went only close to 1.05 V. And that`s where it`s staying.

 

I think your 0.9875V on 11x is pretty good already.

 

Did you hit the default button on the main profile page? If you don't hit it all the multipliers will be the same.

 

i'm thinking about leaving the 11x multiplier at 1.000V just to be safe. if i go below .9875, the system becomes unstable.
though i have stressed the system for 3 hours at .9875 and it came out stable

edit: going to leave the voltage at 1.000V to be safe. Orthos ran into an error at .9875V after a few hours

 

final voltages just in case anyone is interested in comparing:
6x .8500V - 30 min stress test stable
7x .8500V - 15 min stess test stable
8x .8500V - 15 min stress test stable
9x .8875V - 15 min stress test stable
10x .9375V - 15 min stress test stable
11X 1.000V - 3 hours stress test stable

running extended tests now
brought down my max temperature on the 11x multiplier from 90C to 68C!
Thanks for the great tutorial! Computer runs much cooler now with no sacrifice to performance!

 

The profiles were all set to 1.175v for my TL-56 aswell, but I know from CPU-Z that the actual normal voltages are no more than 1.075v so it doesn't correctly detect the defaults for the Turion X2. Try running CPU-Z to see what voltages you are running. For my TL-56, default voltages are:
0.800v for 4x multiplier (800mhz)
1.050v for 8x multiplier (1.6ghz)
1.075v for 9x multiplier (1.8ghz)
They should be the same defaults for your TL-60 with those multipliers, all you need is the 10x multiplier voltage from CPU-Z. I've not seen my power management use any other multipliers (although you can use them with RMClock). You can unselect the other profiles unless you specifically want to use them, but these multiplier voltages give you somewhere to start from.

 

Really Great

By the way is there a way to undervolt the GPU?!

 

I don't really understand. At first, I thought I did, but after seeing CPU-Z's voltages, they threw me off. The voltages that were showing up in CPU-Z were inconsistent from what I set in my profile (following this guide using RMClock). I saw a .800V and I don't recall selecting that voltage. I saw other voltages that I never inputted.

Also, when I selected default button, it did nothing and still kept the all the multipliers at 1.1750V.

I apologize for bringing up more than one problems, hopefully you can hunt them down and address them all.

Thank you everyone!

 

Nope.Sorry.

 

If I got the max temp higher than before during the 45 mins testing, should I stop the test and choose the volt a little higher?

After running an hr testing, should I use the new volt if I got same max temp(71 C) as before undervolting?


Thanks!

 

Yes stop the test. Something isnt setup properly.

What notebook/CPU? Can you post a screenshot of your RMclock profile page?

 

I've found CPU-Z to give incorrect voltages. Compare with RMClocks' voltage shown in the CPU settings page and a 3rd party app if unsure.

For me, CPU-Z never shows less than 1.13v; but my processor will run at down to 1.000v. All other software reflects the lower voltages, and at 1.100v (according to rmclock) my CPU is rock stable for 12.5x (max) multiplier. However, at 1.050v it crashes pretty much instantly.

Thus... CPU-Z is wrong, and RMClock is likely correct. Just check your voltages there (cpu settings, not VID settings - you can sometimes set VID's lower than your processors minimum voltage; it just won't go down futher. For me, I can set VID's to .9500; but my processor never drops below 1.0v.