Simple Undervolting Tutorial - Pentium M Dothan - Page 56

I have a Sonoma powered by a 740. (1.73ghz) And thanks to this guide, I have found a way to keep the laptop slient, cool, and less dusty by undervolting.

Here are my results:
What's interesting is that the vcore goes up by .048v every time my multiplier goes up by 1. And the fan don't turn on at 10x with full load. (w/o undervolting, it turns on when it's at 6x with load)

Oh and I meant to ask, is there any thing that I can do to make the vcore go lower? Cause I know that desktop processors, as they get used, they seem to have more overclocking potential, I know that this is a laptop, but maybe in the same way, we could get lower voltages by simply feeding it with high loads all day. What do you guys think?

need help starting..

Just did a Successful 16 Pipeline Unlock MOd on the XPS Gen1 (3.4 GH non EE, 256MB 9800) and also added a copper mod. 1 Square Copper in at the core of the GPU and 2 rectangular copper shims on the edges.

IDLE temps (CPU=55, GPU=57)
MAX Temps so far seen after 3DM06 (CPU=69, 62)

Since the lappy is my primary business system and runs 9-5 for 5 days, I thought it be a better idea to replace the 3.4GHZ processor with a Pin-Modded 1.7@2.26 Doltan (from my D800 centrino laptop).

My question is:

Would this make my system more stable? I don't use any heavy number crunching progs, but I do have about 43 process running. Would the modded 1.7@2.26 Doltan handle so many processes or would it slow the system (and produce more heat) ?

Can I put the 3.4 GHZ processor (until I sell it and get some money back) in the D800 for the time being. I guess it should atleast run at the max allowed by the Mobo (which I believe is 2.5 GHZ if i am not wrong).

Thanks

following the guide even though its outdated i was able to undervolt my i6000D with pm 725 to .700v at 6x and .908v at 16x and has been prime95 stable overnight for about 10 hours so far. i was amazed that i could go that low that i dont even want to try any lower. what is the lowest anyone has gone??

You can't add a Dothan to your laptop it's a different socket. The Pentium Socket is different than the Pentium 4 socket.

Notebook Hardware Control 1.10 Beta 1 (02.10.2006) Out. Check out: www.pbus-167.com

Can somebody update this on how to use RM clock 2.0 with Prime 95. Also what test did you use on Prime 95. The torture test or the benchmark.
This guide is very confusing.

Hi here is what I did with 2.0

1. install prime and rmc 2.0.
2. run prime 95 torture test - Take the CPU intensive one - top item
3. in RMC - right click on the item in your taskbar and select maximum performance under profiles.
4. go to manage profiles - maximum performance - and now is the important step I found out only after one hour or so - activate use P-state and now go to the bottom and click default - voila now you see the default voltage 1.34 or so.
5. change the default voltage to something reasonable for your CPU like 1.004 - if BSOD restart computer and follow steps 2-5 again - now set something higher. Go so far down until you find the first setting where you get BSOD within the first minute. Now go up two values and let prime run for some hours.
6. go to power saving and drop right down to 0.7 - remember clicking on default. and if no BSOD run prime for some hours.
7. if no more errors in prime - now go to performance on demand - again activate p-states, click on default. then set minimum voltage to 0.7 or other found value, then set max voltage for the highest multiplicator (FID) to your save value and say yes to the question to adjust all other voltages in between automatically.
8. Autostart, minimized, rmc everytime you run the laptop in performance on demand setting.

Voila your done

hello i get an error when i try to boot up my computer saying: NTLDR is Missing. and it says press ctrl + atl + del to reboot and it shows the same warning again. what do i do?

Hey guys I enjoy the lower temps and more battery life, but I have noticed bad performance while gaming with my overclocked X300. Has any of you others noticed this.

...

I've not really caught it yet, but anyway - learning by doing...

@findme: take chkdsk c: /f , you've got probz with your hdd ( mft failed ),
and you sure mean transcend ram, aren't you ?

add: Aaah!!! I've got it ;-D Great advice, thx a lot needledik!

Just wanted to thank you for the nice tutorial...

Got the trick working on my Toshiba M30 with a 1.6 Banian CPU... though the numbers aren't as great as on any Dothan...

Got 6x@0.780 and 16@1.292 ... I'm sure I could get lower than this but I wanna stay safe as i'm also using Clockgen on this so for me it's 6x = 650MHz@0.780v and 1.75GHz@1.292v .... Stable so far ... did some Prime95, SuperPI and some other tests and it never failed ... Although I did get some sound problems before mastering al settings in Clockgen .....

Temperatures still lower than normal (w/o undervolting and overclocking) and performance gain is about 5 secs for 1M SuperPI test and about 280 3DMark03 score points .... Don't know if that's much but thought i'd let anyone using this know .....

I'l do some more tests when i have the time and try and go lower on the voltage .... And i'm going to try and apply some of this on a gentoo linux system .... The undervoltage is fairly simple as you only have to edit a file and input the new voltages and then recompile the kernel ..... but the raising FSB clock is a problem as have no idea where to start... ( a plus for linux is that the system starts up undervoltaged )

Get back to you with final results once I do all the testing ..... and again Thanks ... Great thread ....

PS I can't help thinking that this is not in any way a new discovery for Intel engineers and that they're holding out on us ... Hoping that the Core Duo is as tweak-able as the Dothan ......

Oh and by the way: You can change your CPU in a Centrino notebook and if you upgrade it (the reasonable step) you shouldn't have any heat problems (as previously implied). ......

I use Notebook Hardware Control version 1.10 (v.simple to use) with dynamic switching and managed the following on my Tecra M3.

Multiplier.........Clock Speed.......Stable@NHC 30sec test.......Actually set at
6x..................800MHz..................0.700v*......................0.700v*
8x.................1.06GHz..................0.748v.......................0.780v
9x.................1.20GHz..................0.796v.......................0.828v
10x................1.33GHz..................0.844v.......................0.876v
11x................1.46GHz..................0.876v.......................0.908v
12x................1.60GHz..................0.940v.......................0.972v
13x................1.73GHz..................0.988v.......................1.020v
14x................1.86GHz..................1.036v.......................1.068v

*Unable to test below 0.700v.

I will do more extensive stress testing when I get around to it.

A big thanks to the original poster.

UPDATE: After stress testing voltages are now set 2 increments above stable@NHC 30 sec test instead of 1 increment. Changed voltage details in the above table.

So should I uninstall i9kfangui if I'm running RightMarkCPU clock utility? Also, if I test with the power plug in, I can just use the same values for the battery right? Since the new versions separates AC and Battery for Min/Max Performance.

no, run both... RMClock lets your cpu use lower voltage/less heat for the same processing power...

I9KFanGUI controls the temp when your fans come on... two totally different purposes...

please see RMClock Forum...

btw, CHC is a waste unless you have an ATI GPU...

...

..as I've already said - good idea, great solution, but nothing for gaming...If I play Halo or something, I have to shut down RMClock

NTLDR must be in C:\ and must NOT be compressed even if your C: Drive is...

Finally, undervolting is applied also under Linux! Will try it. Thanks for the link!

Undervolting of ULV processor!

Long time ago I was asking a question about undervolting of ULV processors here.

Now I had an opportunity to do it myself on a 1.2GHz ULV procesor. It runs stable (10hours) at
P-state -- Undervolted - Defaults for ULV
6.0x --- 0.700V ----- 0.812V
8.0x --- 0.700V ----- 0.860V
9.0x --- 0.716V ----- 0.876V
10.0x --- 0.748V ----- 0.892V
11.0x --- 0.780V ----- 0.924V
12.0x --- 0.812V ----- 0.940V

No security margins are included here. It is instructive to compare it with my undervolting results
of the lower level of 2.0GHz PM 755:

6.0 - 0.700
8.0 - 0.700
9.0 - 0.716
10.0 - 0.748
11.0 - 0.780
12.0 - 0.812 (not tested, interpolation between 11.0 and 13.0)

The conclusion is that some undervolted good usual PM processors can serve as good as the undervolted ULV, while not having the higher steps cut out.


Are ther some news on undervolting capability of Yonahs?

the vast majority of ppl don't stress it enough to chk the stability at a given voltage...

i don't consider running just Prime95, no matter how long, to be enough of a test...

i run the Prime95 tests AND while they're running, i run some of my regular apps... i do this for hours, sometimes all day, until i'm completely satisfied it's gonna hold up...

every time i've done this, i've ended up with a voltage that's one step up on RMClock's list, so i just automatically add a step after Prime95 now and i've never had a crash...

I have compared the processor temperature increase after undervolting using RMCLock and NHC
(with the same voltages). The preliminary result is that under NHC the processor runs cooler.

Admittedly, I have to rerun the test under more contolled conditions including the ensuring the identical room temperature in both tests (so far the rmclock test was run in the evening, and NHC in the morning when the room temperature was a couple of degrees (°C) lower).