Edit: I decided to rewrite this into a shorter guide, here. 

I think all Yonah notebooks run on the Intel i945PM chipset. I thought it was impossible to overclock these machines. And over a years span I tried a few times, with no success. But then I came across the Integrated Devtice Technology website. They actually make the PLL chips that begin with ICSxxxxxx. I kind of knew which PLL series were too old and too new to be used on my motherboard, so that let me narrow it down a bit. I downloaded the datasheet for ICS954204 (attached). I think ICS makes them for whoever uses the PLLs, like BIOS programmers I guess. But it turns out they're not very hard to read, at least for our purposes. Pretty much we're looking for something that deals with the FSB clock right? So, that's not hard to find, it's actually layed out right on the first page in a table:

Okay, looking at this alone doesn't help us figure out how to use it (it's much clearer in the ICS954321 datasheet though). But if we keeping reading through the datasheet, or if you search for FS_C, B, or A, we'll come across Byte 6. There we can see that the first table is describing Bits 2, 1 and 0 of Byte 6.

Well if we look for the familiar (what our notebooks' factory settings are) we'll see it in row 4 (first table). Row 4 (166.67MHz FSB) reads Byte 6, bit 2 = 0 | Byte 6, bit 1 = 1 | Byte 6, bit 0 = 1 and that agrees with our IC dump from ClockGen:

So naturally, with overclocking in mind, we want to change Byte 6 to the values of row 1, 5 or 7! Starting with the lowest overclock (row 1) I set the Binary Value to 00101000, hit Update Selection, and Apply Selection. At this point my laptop googled for the child-abuse hotline and locked up. The result is the same for the other overclocks. So this route was not going to work for me, but it might for you.

Reading on, we come across another description of Byte 6. It looks like Test Mode is set by default (bit 7 & 6). Well, maybe turning off test mode (changing Byte 1 to 7F) will actually let us use the FSB slider. No, strange that doesn't work either.

Reading further along, we'll find Byte 9 (page 13). The name alone seems worth looking into "Spread Spectrum Control." While looking at another datasheet (ICS954321) I saw that the spread spectrum was a way to change the clock in smaller increments, unlike the massive FSB changes found in the Byte 6 trick I tried in the beginning. I actually don't know what a Spread Spectrum is, but I do know it's directly related to the FSB. And our overloads, I mean the scientists, have given it a name that sounds like it relates the FSB clock. Seeing that there was actually a Control Register for it in the ICE954204 datasheet was promising. The thing about it is that we can change the Spread Spectrum values (bits 7-4) but nothing happens. All it does is change the size of each increment if we were to actually use the Spread Spectrum.

But look at bit 0. It's a curiosity. Reserved and set at 0, I wonder what would happen if we flip it to 1. Why not? Let's see what happens. Well, it would appear nothing happens. But in fact, something did. If we try to change the FSB with the slider the changes will now take effect. That reserved bit is actually a switch to turn the SS Control on or off.

This writeup isn't an exact retracing of my steps. IRL, once I found what what ClockGen and SetFSB really did (modify the PLL registers) it took a couple of weeks of abusing my laptop and making guesses at what to change. But in summary, if you want to overclock your Yonah (or maybe even any locked FSB board) find the Spread Spectrum Control Register and the bit that will enable it. In i945PM chipset users' case this is Byte 9, bit 0. Set Byte 9, Bit 0 to 1 and slide the FSB to your desired value.

The PLL doesn't have to be an exact match, I see somethings in my IC dumps that don't match the defaults for this datasheet we used. But it's close enough. I have an Acer 5672WLMi x1600M. I can use the ICS954123, ICS954148 and the ICS954227CGLF. I believe there were one or two others that worked, but I don't feel ike testing them right now. But you can see the basic range of PLLs for the i945PM chipset is ICS95312x - ICS95xxx (or something like that).

205x10MHz is the highest stable overclock I could get with my laptop (I hit 2077Mhz, but I couldn't even run SuperPi with that). That was with some G.Skill PC2-5300 that was running at over PC2-6400 specs. I thought if I got some real G.Skill PC2-640 I could kick things up a notch, but ironically my maxium overclock is now 203x10Mhz. That maximum is completely stable however. I'll update if I can figure out how to squeeze more speed outta this sticks.

ics954321.pdf (224.64 kb)