I'm not sure exactly what you're asking for... the mobile Bartons were never locked to begin with.

Are you wanting to change the default multiplier?

Ah, from reading your response, I see that my O.P. contains ambiguous and confusing information. Yes of course you're right - the XP-Ms are factory unlocked, but the interesting thing is, if you override their maximum multiplier value by messing about with the L6 bridge, you can increase this maximum multiplier value. For example, the 2400+ rated mobile barton has a maximum multiplier of 13.5x, but if you cut in to the L6 bridge with a knife, that goes up to 18x. I'm not sure if the term "maximum multiplier" equates to "default multiplier", as you say in your post, because I'm unsure about the correct terminology. I suspect that these phrases are technically the same thing. So, yes that's what I want to do - I want to be able to increase the default multiplier on this particular 2600+ rated mobile barton.

That info link you sent looks like what you want to do to unlock a regular Barton, not a factory unlocked mobile chip.

bushwack wrote:

That info link you sent looks like what you want to do to unlock a regular Barton, not a factory unlocked mobile chip.

It's funny how a tiny piece of formatting can mess stuff up. That link I provided had a . (period / full stop) punctuation mark at the end. This prevented the .html#L6 link from working correctly. I have now removed the . symbol, and the link now takes you to the correct image on that fab51 website. However, it's exactly the same image that appears in the O.P. anyway. And I do believe that this table contains information about unlocked mobile barton CPUs, because the CPU IDs listed in the table match up exactly with the XP-M chips seen on cpu-world .com

Ah, I see now.

But if you have a 15X multi already, why not just set it to 16X or 17X to start with? Either one of those still only requires cutting one bridge, same as the 22X setting... Unless I'm overlooking something here, it seems pretty straightforward.

Why do you need such a high multiplier? You going to be running yours at 3ghz??? I figure the minimum bus you would want to run a barton at is 200.

The only reservation I would have with using the 22x settings is how boards with AMI BIOS will start with that setting. 2.3ghz is usually the wall on those processors, so in that situation POST'ing with an FSB higher then 100mhz is fat chance.

Your 8KTA3, using the Award BIOS, won't have that issue. I guess for testing purposes it would be cool, if you interested in seeing how 100mhz vs whatever scores/compare. So unless that is case, 17x setting would be my choice.

@ prophase_j - What are you running your XP-M at? I ran mine at 11x200 for years.

IIRC it would run stable @ 2400, but that extra 200mhz really pushed the temps up, made my machine audible, and was only just a few percent faster in gaming.

I have a 2500+ running a 166mhz bus for a 2324 total. To do this it needs a bump to 1.725 and runs pretty hot. I also have a "real deal" 3200+ on an nforce2, which has been tested close to 2.4, but with a solid copper heatsink and 1.875 volts. I'm saying it's impossible, I'm just not into testing the limits of power supply or heat dissipation 😀 Besides, if your really into clock speed you should have a Thoroughbred anyway.

You've all made me see the light. 17x multi is the obvious choice. Thanks guys. All I need to do is cut L6 bridge, section number 2.

I have an untampered XP-M 2600+ here. Just for you, I changed the clock to 16x133. No problem! I don't know why you want to cut bridges. The CPU can run faster than 15x without cuts.

Perhaps the problem is with the mobo / bios?

ux-3 wrote:

I have an untampered XP-M 2600+ here. Just for you, I changed the clock to 16x133. No problem! I don't know why you want to cut bridges. The CPU can run faster than 15x without cuts.

Perhaps the problem is with the mobo / bios?

Awesome! Please can you tell me how you adjusted the CPU's multiplier from its default setting of 15x, to the new setting of 16x? Did you do this inside the mobo's BIOS set up area, or did you use a software utility? What is the model name of your mobo?

My Epox mobo's BIOS does not allow me to select a multiplier higher than 14x. Also, inside Windows 98 SE, I am using the utility called CPUMSR.exe. When I run it, I cannot increase the multi to beyond 15x. This is because the 2600+ rated mobile barton is currently untampered.

Also, does your 2600+ XP-M run at 133 mhz (266 DDR), or at 166 mhz (333 DDR)? The reason I ask is that cpu-world .com tells me that there are two different bus speed versions of the 2600+ chip.

Edit: Apologies. I now understand. You must have this XP-M:

http://www.cpu-world.com/CPUs/K7/AMD-Mobile%2 … A2600FKT3C.html

It's a T-Bred, with a default multi of 16x. That explains everything. My XP-M is a Barton, with a default multi of 15x.

I used the bios from Epox 8-KRAI. Based on KT-600, no V5. However, one can put in a GF6800 and use a glide wrapper. Gives AA as well, but doesn't work on all games.

My XP-M defaults to 133*15. But I run it at 200*11 basically from the beginning. And I just tried 16.5x133 with success.

I have thought about moving it onto my Kinetiz 7A, but can't be bothered any more. Virtually any game in need of more than 1400 MHz (Athlon TB) will work nicely on my current C2Q system. The Kinetiz 7A would be living on borrowed time anyway, thanks to the cap's limited lifetime.

I've declared peace on the universe and settled for a slot1 BX machine with a P3 at 1000 MHz or a mulit-limited P2. I managed to install an 8 inch fan independent of the cpu, so that I can switch CPU in less than a minute (Pull and Plug). And since that decision is off my chest, I feel better.

ux-3 wrote:

I've declared peace on the universe and settled for a slot1 BX machine ...

😀

I've run into kinda the same problem regarding the barton-m multi's. Sometimes they will be stuck at 6x in the BIOS. The limited max multi I find is less of a problem as I will usually overclock with FSB anyway.
I haven't really tampered with those 'more-locked' XP-M's really, I'll just do some reading about them. Anyway, I think crystal-cpu may solve this little multi-problem I have with them.

I've just had time to "operate" on my XP-M 2600+ CPU. Recap - its default multi is 15x, and FSB is 133 (266 DDR). I cut L6 bridge number 2. This means that its default multi has changed from 15 to 17. The cut operation was successful. I used a sewing needle, held tightly in a pair of pliers. I just pushed the needle down in to the CPU surface. I got lucky - the cut was in the right place, at the right depth.

I have come to the conclusion that increasing the mobo's FSB is more profitable than increasing the CPU's default multi, in terms of a performance increase. For example, I can increase my Epox mobo's board to 152 FSB, without it complaining. It can run either an AGP or a PCI Voodoo5 at that FSB setting. If I then increase the XP-M's multi from 15 to 16, the system becomes unstable. If I reduce the FSB value to below 152, I can successfully increase the XP-M's multi, but I don't get the same performance increase. I think that increasing the CPU's default multi is only worth doing, if for whatever reason you are unable to successfully increase the mobo's FSB.

Thanks a lot for all the advice, it was a fun project. 😀

Before I move on to other things, I wanted to retest the 2600+ rated (2 GHz real) XP-M CPU. After half a dozen tests, it seems that the "specially unlocked" XP-M is able to tolerate a multi increase from 15 to 17, if the FSB is not increased beyond 140 FSB. This gives me the opportunity to overclock both the mobo's FSB, and also the CPU's multiplier - just for a bit of fun.

The image below shows CPUMSR.exe, 3DMark 99 Max, and Sandra 2002 Pro. CPUMSR shows the multi and CPU clock speed increase. 3DMark shows a Voodoo5 5500 PCI card in action, and Sandra shows the FSB increase.

I decided to remove the 2600+ rated XP-M, and drop in a 3000+ rated XP-M. This thing runs at a real clock speed of 2.2 GHz. I am reluctant to "operate" on this processor, because I can see the benefit of simply overclocking the mobo's FSB only, to maximise system performance and stability. However, it is still necessary to use CPUMSR and its low level companion driver to adjust this CPU's multi inside Windows, because the Epox mobo's BIOS does not "understand" the CPU's default multi, which is 16.5x.

The image below is similar to the previous one, and should be self explanatory. I'm still using a Voodoo5 5500 PCI card. I know for a fact that the V5 AGP version is faster because it scores about 11800 for the 3DMarks, but I left the PCI version in the mobo, as I was curious to see if it was stable with an OC'd FSB. It seemed to cope OK.

I have made a couple of changes to the system:

* Removed the PCI V5, and replaced it with an AGP V5.
* Overclocked the AGP V5 by 10 MHz, from 166 to 176.

The image below should be self explanatory.