how to identify voltage regulator to undervolt?

i'm looking for the voltage regulator to in order to hardware undervolt it. what general area and numbers should i look for to identify it? the notebook is a asus UL20A if that helps.

 

Why not just pin mod the CPU to undervolt?
http://forum.notebookreview.com/showthread.php?t=390696

Also while you have it open, check what PLL you have:
http://forum.notebookreview.com/showthread.php?t=429383

 

i don't think it's possible to pin mod because i can't take the CPU off. it's a SU7300 ULV chip, which i think is soldered on the motherboard. what's the difference with pin mod the CPU and switching around some of the voltage regulator stuff?

and it seems like the clock generator is ICS9LPRS365BGLF, same as the acer 1410/1810t. i tried to use setFSB to overclock, but it locks if i up the fsb just by 1mhz. this means i must do the TME mod also right?

 

Did you try software undervolting first? Also, the clock generator =! voltage regulator. Not to mention voltage regulators are going to have several different software-controlled voltage settings, so I don't think a pinmod is going to do what you think it is here.

I could be wrong. In fact, I often am, so feel free to correct me.

 

yes i tried RMclock to undervolt it, but it doesn't go as low as i want it to. hardware undervolting is a bit harder to do, but it will let me undervolt lower and doesn't require me to have windows automatically load RMclock on every boot.

ISL6261 is the voltage regulator on the acer 1410/1810t and i was looking for something similar, but i can't find it on this asus board

 

You could try changing the dsdt code in your registry to undervolt. Means not having to run RMclock.

Also about the PLL, looks like you dont have TME mode enabled.

The reason for the lockup is probably because setfsb is changing something that it shouldn't.

For me it was the source for the SATA clock.

So here is what you can try:
1. Press getfsb.
2. Click on the diagnosis tab.
3. move the top slider.
4. press setfsb.
5. Use a camera to take a photo before the notebook restarts.
6. log back into windows.
7. start setfsb.
8. click getfsb.
9. click on the diagnosis tab.
10. Change each control register one by one using your photo until the notebook locks. Remember which byte made it lock.
11. restart and start changing the bytes again, this time not changing the byte that caused it to lock last time.
12. Then start changing the bits in the byte that caused the lockup (until you get the lockup).
13. restart again, start changing the bytes and bits but this time leave out the bit that caused the lockup.

If the overclock doesn't make the notebook lockup this time, email Abo (creator of setfsb) and tell him which bit causes the lockup. He should give you a test version of setfsb.

 

cool, much thanks moral hazard. i will try that once i piece back my notebook lol. it's in so many parts right now.

would it help anyone locate any of the voltage regulator components if i took pics and posted them here? i don't think my camera has a high enough resolution to read all the fine prints though

 

ok i'm ready to try this. which values do i need to change in my registry?

 

First download the attached file.
(you will need it to modify the ACPI stuff).

Then also download this:
http://www.acpi.info/spec.htm

And now comes the part I have never done before (the code changing to undervolt).
Though I have messed around with the dsdt a lot, I havn't undervolted with it, so give me a min to find what value to change.

 

You will be doing something similar to this:
http://forum.notebookreview.com/showpost.php?p=5627558&postcount=31

So anyway, follow the guide I attached to get your ASL file.

Please upload your ASL file here so I can see.

 

ok i'm going through the steps so far, and i'm stuck on which one is the "subkey" to copy. is it that 0000000 number? how do i copy it?



EDIT: nm, i think i got it. this is the subkey path, but how do i paste in the command window? i'm guessing i just have to type it all out since i can't do Ctrl + V to paste in there

HKEY_LOCAL_MACHINE\HARDWARE\ACPI\DSDT\UL20A\UL20A205\00000205

EDIT2: nm, i got through all the steps to get the ASL file now. will upload now...

 

I don't think I know exactly what to change in your file.

I'm going to PM Darth_nVader and see if he can help out on this thread.

 

moral_hazard, i greatly appreciate your help so far. i'm surprised no one has every done this before. hopefully this can come out successful so others can try on different notebooks as well.

btw, where are the fan speed settings? i can't even find anything between 0x64 and 0x0.

 

Your dsdt does have the fan code, but it's been programmed in a different way.

You could probably figure out how to change the fan speeds, but again I could'nt really help there.

I did PM Darth_nVader so you might get some help from somebody who knows.

 

While you can add P-States to your DSDT, most of the time they are defined in one of your SSDT's.

Dump and decompile, and search for:

Code:

_PSS

.

I'm not sure how to override SSDT values in Windows.

 

You said you undervolted the SU7300 with RMClock?

How did you do it? It doesn't seem to work for me. I can only play around with multiplier 6.0x, which is not sufficient for this CPU. So it seems i get wrong readings in rmclock, but why?
64 bit driver installed, set the unlockvid dword accordingly... do you have any tips?

 

While I have been able to edit my DSDT for Lower Voltages at Higher Multipliers, I can't go below my Boot_VID.

My lowest P-State 0613, I tried to write a lower value(0611) to MSR 0x199 but when I read back MSR 0x198 it's still 0613.

While there does not seem to be a VID_MIN, the voltage from the VID_BOOT is as low as mine will go, without entering a C-State.

When one core of a Dual Core CPU enters a C-State your voltage should drop by around 1/2.

If you want to lower the Voltage of your CPU at it's MAX ratio or one of the intermediate P-States, I can help with that, however I can't say how to do that for the Lowest P-State( I can tell you How in theory, it just won't work for me.)

 

An imperfect solution is still much better than none at all. I'd be happy with any power saving, as little as it may be.

I am not an expert on this subject, so i better don't go out and just try myself. Any input would be greatly appreciated!
Thanks in advance, mate!

 

yes i undervolted with rmclock. you have to unlock some stuff in the registry and it will let you select lower voltage values, however, the lowest i'm able to go is 0.90v. just keep in mind that rmclock can't do half multipliers. the SU7300 is naturally 6.5x200mhz to get 1.3ghz, but rmclock will only run it at 6x200mhz for 1.2ghz. that's why hardware mods can overcome these problems.

sure, this is also helpful and better than nothing. how do i drop the max voltage below 0.90v? my notebook won't be idle the whole time it's on, so this will help reduce heat and increase battery.

 

Use MSR Tool to read the MSR 0x198 and post the result.

The values from that will let us know the Valid Range.

 

sorry i'm a little n00b, but how do i use MSR tool? i DL'ed it and opened the program, but nothing happened

EDIT: nm, i just found out i had to run it as administrator. here's the screen shot of the 0x198

 

On a Core 2 Mobile CPU ( Your ULV may differ ).
0x0F = 0.9125v
0x10 = 0.9250v
0x11 = 0.9375v
0x12 = 0.9500v
0x13 = 0.9625v
0x15 = 0.9750v
0x16 = 0.9875v
0x17 = 1.0000v
0x18 = 1.0125v

Or very near that, as Intel does not Document it without NDA.

If you disable SpeedStep, you can test these with MSR Tool.

MRS Number 0x199->Click Read MSR

Then Click the Write button at the TOP(Core 0), that will fill the Write MSR fields.

Edit the Write MSR Fields:

0x00000000 0x0000460F->Click Write MSR

REPEAT for Core 1 ( Don't forget )

Click the - Button and it will read MSR 198 it should look like:

0x060F4618 0x0600 460F

Run whatever stress test to see if it's stable, if 0x0F is not Stable goto 0x10 and so on, until you find the lowest stable Voltage.

What OS or OS's are you running?

 

Darth nVader, thanks it worked like a charm. cpuz idle reads 0.875v and while stressed, it reads 0.90v. is there a glitch in CPUz that when idle, the multiplier is 8? the max for the SU7300 is suppose to be 6.5, which is seen when it's stressed





everything's good so far, HOWEVER, here's the few problems i ran into. i can't disable speedstep in the bios, so i can only set MSR correctly while the CPU is stressed, otherwise one core can be set. setting one core will cause the other one to return back to original setting.

one other problem is when after i stop the stress testing and the voltage drops, the MSR values that i set before were not retained and the voltage shoots back up to the original 1.013v. i think is because of the speedstep right? i need to disable it, but how do i do that if there's no option in the bios?

my OS is win7 home premium 64bit if that helps. btw, the bios has an option to have a 5% overclock, so you can that in the fsb. since the bios has this option, is there a way to write the DSDT table to increase to a higher overclock? if i can't undervolt any further, then might as well try to overclock with the voltages i have

 

Disabling Speedstep was only for testing, it's better to leave it on.

I'll have to do some research, you want to OverRide the P-State values your Bios hand off to Windows 7's OS Power Management.

That way you don't lose SpeedStep, and it just works, without having to change anything after each reboot.

I'll look into it.

You can do SMBUS Clock Generator control in a DSDT, however it's not easy, there are better ways.