The "Undervolting" Guide

It sounds to me like the system drivers from Lenovo are not loaded, or are not loaded properly. If they are, then you may have systen board problems. You can download the bootable PC-DOC image from the Lenovo site, burn it to a cd, then run a full system board diagnosis.

I do not know the details of the system board, so I can not comment on the sensors location.

 

thank you, stress testing orthos now on my 1640

 

Um, I'm not on a Lenovo..I'm on an Acer.. The T9600 and T5800 are the processors I had, oops sorry, I think I didn't clear that out. I think you thought I was talking about Thinkpads.

Anyway, are these system board diagnostics programs applicable for all laptops?

 

All Dell's have system board diagnostics, but it needs no CD. It's an option in the [F8] boot menu.

~Ibrahim~

 

Okay, we are all straightened out.

I would check the Acer support site and their forum.

I think the Ultimate Boot CD has tools to run basic tests on system boards, but I do not know if they will detect mal/non-fuctioning thermal sensors. The disc is a free download, so it can not hurt to try.

There are third-party commercial programs also.

 

Well, I just leave cpu-z on and then do my stuff. I can see the frequency change, but my sound doesn't click. If I set it to "Performance on Demand", the frequency still changes but the sound will click constantly. Since I can run my P8400 at 0.85V for any speed, I made a profile that locks the cpu to the highest speed when on AC. I gain a couple degrees in temp but no more annoying clicks.

 

I'd like to know if, in rmclock, I've to check or not the "use throttling" option.

Currently I've not checked it but the cpu, however, scales correctly.

 

dont use throttling!

It ruins CPU performence
But it will reduce temps.

 

I've done some experimenting on my 1810t:

 

nice work there.
Overclock + undervolt?

 

Yep
I expect su3500 to be able to reach 2GHz with proper RAM.
0.875V is the SuperLFM voltage

 

Throttling doesn't affect frequency or voltage. It just parks the CPU in idle for a percentage of the duty cycle. Speedswitching is different. Undervolting has absolutely nothing to do with the above two items.

 

Hmm there's something I noticed a while ago.. I switched back my mom's laptop from the T5800 back to the T2370, and I noticed some components in common with GPU's.. I wasn't able to take pics of the CPU's but I saw these near the processor socket as well: (See photo)

The green one, I saw one of those under the very socket of the processor. Kinda sits a little bit to the side, kinda touches the underside of the core. The blue one, I saw two of them a bit to the left of the socket--maybe about a centimeter away. Do you guys think these are the thermal sensors or something?

 

rmclock doesn't show the advanced cpu settings tab for me D=!

 

Throttling does not change voltage or clock speed. It simply parks your CPU in idle for a portion of the processor duty cycle. Unless you want to turn a 2GHz laptop into a 1GHz machine to improve your temps score in the UV "game", I wouldn't use it.

 

RM clock can't recognize my cpu and i can't do anything with it.



help please =/

 

Anyone have experience with the issue below?

If RMClock is turned on while hibernating, the computer BSODs with an ATI dll file error. BSOD screenshot: http://img.photobucket.com/albums/v58/trieper/IMG_1717_BSODonhibernatingwithRMClo.jpg

Hardware: Asus F6Ve, CPU Intel T9550, GPU ATI Radeon Mobile 4570. I'm using ASUS-released GPU drivers (I don't think ATI has released mobile drivers for this card).

Tried reinstalling the driver and also the ASUS power management software but that didn't solve the issue.

Thanks in advance for any info.

 

Using Windows 7, have RMclock startup, but everytime I start up Windows, the UAC pops up. Anyway to disable it without disabling UAC totally? Thanks.

 

I have a strange problem. My computer is LG E500-SP16V and the processor is T7100 and I'm running Windows 7 x64. After waking up from sleep my computer just freezes. I found out that this is caused by DFFS. If I disable it from settings, computer wakes up normally, but naturally I can't use SuperLFM anymore and even the lowest voltage raised from 0.8500 to 0.9375. I tried to flash my BIOS to older version without any luck.

So I'm just wondering is this a known issue or is my hardware just broken somehow or something else? I could just stop using sleep mode, but it's so handy. Well, maybe I just disable DFFS.

EDIT: I did some further testing and it seems that I can enable DFFS from Advanced CPU Settings, but I can't enable SuperLFM from profile section. When I do that it freezes immediately.

 

Keep S3 sleep. Ditch SLFM. It doesn't serve a real purpose. Use S3 standby to save battery and top FID when you want to work. Prove it to yourself.

 

Yeah, I'll do that. Now that I can after all enable DFFS and use voltages as low as 0.8500V, lack of SuperLFM isn't anymore a big problem.

 

What, exactly, do you use your computer for?

 

Yea, I double checked that, it's set right - even without rmclock installed, stock voltage is 1.625 across the board with small runs at 1v or 1.238 depending on load. maybe different processor batches or something like that I guess. 0.925 is the lowest my cpu will try to run at.

 

Is not throttling the "same" of scaling ?

 

I don't understand, can you rephrase that?

 

Sorry, don't throttling and scaling have the same meaning ?

There are also two other options in the 'advanced tab' of rmclock: thermal throttling and normal throttling.
Have I to check them or not ?

 

If you would like your CPU to be normal, uncheck both throttling options.

But if you have overheating problems, I suggest you only check "normal throttling".

 

Which are the consequences of selecting 'normal throttling' ?

 

RMclock gives me only 2 indices for my su3500 in the profiles tab, with resp. FID 6x and 7x, is that normal?

 

gs45 isn't fully supported yet. 6x and 7x is normal. Check Timeline Tweaks for a tutorial how to undervolt su3500.

 

your post #87 on page 9, I presume that is?
thanks indeed.

 

There is a new Timeline Tweaks thread(see signature). There is a link to this tweak in the first post called "Optimizing battery life and undervolting".

 

Just started using this undervolting technique...with my HP dv5t, I have an Intel T9400 2.53 Ghz.

So far I have
0 - SuperLFM - 0.9000 v
1 - Normal - 0.9000 v
2 - Normal - 0.9125 v
3 - Normal - 0.9125 v
4 - Normal - 0.9875 v

Max temp I'm getting is 78 degree celcius. Anyone with the same model or at least same CPU? What are your settings? My notebooks mobo is the PM45.

Also, at startup, on Windows 7, I always get the UAC message, is there anyway to disable it WITHOUT having to disable UAC completely?

 

RMClock can't disable overheat throttling, though? The throttling where if the CPU gets too hot, it automatically underclocks and undervolts itself to save itself from dying, right?

~Ibrahim~

@sendmarksmail

http://www.mydigitallife.info/2007/...ac-or-disable-elevation-prompt-with-tweakuac/

 

Slower CPU.

If you are playing a game, you will see the frames per second go down.
Generally the CPU will not be working as hard as it should be.

 

Below should be in the first post:

 

I also use a program called "Startupunblocker" that is just one executable, and you can select among the programs in your startup folder, user registry and system registry startups and edit their startup parameters. I found this to be easier since I use this to execute Launch Manager as an admin at startup to make one of my shortcuts there work. I found the Task Scheduler a bit unsure, since sometimes it stops the task even if I tell it not to.

 

Couple of questions:

1) If I am going to be using a notebook cooler most of the time, should I perform the stress test while my laptop is on it?

2) Currently I am in an air conditioned room with a ceiling fan. When I take this elsewhere, it may be warmer. Do I need to account for this?

3) Is AMD Turion X2 going to be much different direction-wise?

Just stressed it: 64 c min and 84 c max, not bad I think considering this laptop has been powered on pretty much 96 hours straight. I will restress tomorrow so that I can have an accurate reading at that time as I do not have time to run the next 45 min test now. Check my sig for my rig, and I will try to remember to my steps into a post on here for others to follow. Cheers.

 

Yeah, you're right there.

You would have to modify your dsdt table in your regisrty to stop that.
I can give you a great guide + software for that if you want, also it may allow you to control your fan speeds and the critical trip point.

 

I came to the same result as der mali with index 1 (FID7) in RM, i.e. .875V.
#0 (FID 6) put at .725V in the profiles- as well as in the subprofile-tab results under load (orthos) in a shown VID @ .875V for FID 6.
Maybe I miss something?

 

Like I posted in that tutorial, going lower than SuperLFM voltage doesn't work. RMclock lets you select lower voltages but they have no effect. 0.875V is the lowest possible value.

 

thanx for the explanation.

 

I did some tests with a HP DV4T with T5800 CPU. The laptop was plugged into a Kill-a-watt electricity usage monitor. The Kill-a-watt measures power draw to the nearest watt, so it's not high precision, but sufficient for some simple sanity checks. Here are a few observations:

- With the CPU and disk idle, the display at med brightness, and wifi off, the power draw was about 17 watts.

- With the CPU and disk idle, and a constant VID of 1.050, I could not see a difference in power draw between the lowest 6x and highest 10x multipliers. I also could not see an difference in the CPU temp. This seems to confirm what Scott said earlier, that speed stepping in itself does not conserve power.

- With the CPU at full load and a constant VID, the 10x multiplier drew about 7 watts more than 6x. 10x definitely had higher CPU temp than 6x.

- The difference in power draw between lowest and highest screen brightness is 4-5 watts. (standard CCFL backlit screen, not LED.)

 

SuperLFM is a joke for me too. I mean, i have a wattmeter, and the consumption/cpu temps remain EXACTLY the same between superLFM 6x and 6x.

 

Even with a cooler you should still undervolt your CPU. The cooler the notebook is running, the longer it will last. I undervolted mine using K10STAT, found here, yesterday and have seen good results. I set P-State 0 to 1.0625, P-State 1 to .9000, and P-State 2 to .7750 and under stress the processor now goes up to 78C max. On my cooler, 75C is the max. When I am not gaming and just web-browsing, 51C is the max it has gone without a cooler.

 

I have a Dell XPS M1530, with T9300 processor. I followed the instructions on the first post of this thread to the letter then tried copying settings from Undervolting M1530 (T9300) - Worse performance? thread.

However on testing I don't see any improvement in my temperatures, my CPU still pushes just over 90degrees C. Could there be anything obvious I'm doing wrong why it would not work?

 

I would suggest finding your own minimum volts.
also if you could download cpuz (latest version) and check if your voltage was really changed.

 

I tried finding my own first, didn't notice any difference in temps. Will download cpuz.

 

Gorgeous, just got my t7300 (sig) down to 0.900V stable (2hrs orthos, 1hr gaming), and my portege r100's pentium M down by about 0.200 at max and 0.200 at minimum - gained nearly 1hr battery life.

Eight hours. From a 12.1in laptop. Who needs netbooks?

 

Exactly !

Undervolting saves power across the board. Once you eliminate any possibility of over heating by UVing, speedswitching and throttling become irrelevant toys.

OK, I broke down and bought a data logging power meter too. Geez. Next project is to show (again) that power conservation really only depends on load management. Anyone is welcome to try it.

Holding all things constant (display, HDD, LAN, etc) measure watts and time for a fixed Super-Pi job such as 32M at two different FIDs. There is a simple mathematical relationship that explains the results: At identical voltages, twice as fast for half the time requires the same mW-Hrs of juice as slow speed for the regular time.

If your top FID voltage is higher (usually is), then you pay a premium for faster processing. But once you're back at idle, where your CPU spends most of the day, the FID setting is again irrelevant as far as power draw. Isn't it?