KX7 with multi above 12.5

mmonnin

I have a KX7 with a CPU in it that does 2.3+. A 1800+ DLT3C JIUHB. The thing is its not stable at anything over 173 at the 12.5 multi. Since it doesnt have a 6:2:1 divider I cant go much higher on the FSB so I was wondering if there are BIOSs out there that have mulipliers above 12.5. The CPU can handle it. RAM is PC3200 so thats ok.

Does anyone know. I still have the BIOS it came with. No use to upgrade if it works until now:)

I could use some more MHz for FAH:):fold:

If there are no BIOSs with this option then I guess I could try Pin modding the CPU to 13 or 13.5

Jengo

well, i bought this KX7 from seversphere, i am almost positive it has the newest bios revision, problem is, i dont know how to flash, or how to give it to you, heh heh... (never bothered to learn how)

im not sure how high the mutliplyer goes on this one (sorry im a bad OC'er) i dont remember, i will post back later with the answer

heh heh...

muddocktor

Marc, AFAIK, the KX7 won't let you select a multi higher than 12.5 in bios, just like most other boards from that era. However, you can pinmod a higher multi or do the L3 mods for a higher multi and it should work fine by setting 12.5 or higher in bios. Your present L3 configuration should look like this for the 1800- : l l l l . The easiest L3 mod for a higher multi for you would be to change it to a 15.5 multi, which would look like this- : l l : : , with you cutting or blowing the last 2 L3 bridges. That would be a good multi to run PC2100 memory on. Another choice for you would also be the 13.5 multi, which looks like this - : l : l : . This would be better for PC2700 memory as you could run a 166 fsb speed or even a little higher without running into pci/agp problems.

As far as the bios is concerned, check Abit's site and see in the bios descriptions if it enables softmenu selection of higher multi's with the latest bios release, but I don't think it does.

mmonnin

Heck it froze or something last night at a 170 FSB. I thought thats what I ran it at before but its not stable.

The 15.5 multi would work. I have a stick of PC2100 that goes upto 156 MHz. That would go upto 2.5+ GHz, prolly too far for the CPU but I could still get the max then.

Spinner

muddocktor wrote:

As far as the bios is concerned, check Abit's site and see in the bios descriptions if it enables softmenu selection of higher multi's with the latest bios release, but I don't think it does.

Yes, the latest revision doesn't have anything above 12.5. My mates got one, revision 1.1 of the board, and I've been reading up about them because I wanted to put a Barton in there. Which of course runs great. But yeah, obviously no 6:2:1 divider (ABIT officially doesn't even make use of the 5:2:1 divider, hence why they rufuse to add official support for 333FSB CPU's even though they do work!), and nothing above 12.5 in the BIOS.

Cheers

lsevald

I thought it was possible to switch the low(<13x)/high(>12.5x) multi range by modding the last L3 bridge (+8x bit). The KX7 only supports a 4bit FID (+0.5x, +1x, +2x and +4x), so by manually setting the 5th bit to high (open bridge I believe), the lower multi settings in BIOS (5-12.5x) will be translated by the chip to the higher range. So for instance if you select 5x in BIOS, you will get say 14x (or some other multi in the high range, only guessing here). Or did this trick only work on the Palominos?

mmonnin

I remember something about what you mean lsevald but I dont remember what it applies to.

There is like a <13x or something similar option in the BIOS but it doesnt work at any other settings.

Wish there was a 6:2:1 divider. Then I could do 200 FSB and still be at stock AGP and PCI frequencies.

muddocktor

Lasse, I believe that will work but since he already knows what that Tbred can overclock to, you might as well just go ahead and hard code a 15-17 multi on it anyways, since you would have to mod the last L3 bridge to do that. Any settings that are reserved will still stay reserved either way.


BTW Marc, don't go for the 15.5 multi as I now noticed that it's labeled as reserved and won't work. A 16.5 multi would be easy to mod to also, with the L3 config looking like this: : : l : : . That would give you a default speed of 2195 on a 133 bus, which should be doable if the proc is good for 2.3 already. If you don't mind connecting that factory cut first L3 bridge, then you can set whatever multi you want on the proc or you can go through the trouble of doing the wire in the socket trick or wire wrapping the pins on the proc also.

mmonnin

Actually I tried something out and my other 1800+ thats in my second NF7, I upped the multi to 12 and its running at 2.4 I just dont think I gave it enough voltage before. So most likely they can both do that.

I would like to try something that can be undone. But if it gives me the option of just running a lower FSB then I could still move it to another board if needed.

I tried the wire trick and it was a bitch.

lsevald

muddocktor wrote:

Lasse, I believe that will work but since he already knows what that Tbred can overclock to, you might as well just go ahead and hard code a 15-17 multi on it anyways, since you would have to mod the last L3 bridge to do that. Any settings that are reserved will still stay reserved either way.

Agreed If it's the old type with hidden traces that is. The new types with visible traces are just too hard to fully bridge mod in my opinion. I did two unlocked week 31 2500+ CPU's a week ago, and I spent the better part of two days to make them cooperate at 16x on an Asus A7M266-D. Actually I wasn't aiming for 16x (I wanted 18x), but as I was unable to blow the first and third L3 on one and the third on the other, I had to settle for that and get the rest by upping the FSB. On the one I managed to blow the first L3, there is now a big black pit where the bridge used to be...it took that much to disable the darn thing. Funny that the first and third L3 were problematic on both chips

mmonnin

I have 2 week 21 1800s.