confused about L3 bridge cutting

[10GHZ]

based on http://fab51.fc2web.com/pc/barton/athlon-e23.html

i have a Xp2400+. Do i just join the 5th L3 bridge to enable the lower multiplier? if thats the case, does that mean once i've connected the bridge, in my bios now 7x really means 15x and if i up it to 8x, i'll get 16x multiplier on my 2400?? thanx

 

[phsstpok]

That link is very confusing. Why do they they have diagram upside down?

Anyway, you've got it backwards. Connecting the 5th L3 bridge will force the 8x bit LOW. This will only give you access to low multipliers 5x - 12.5x with <b>NO REMAPPING</b>.

If you "cut" 5th L3 bridge (yours is already cut) you have access to the higher multipliers.

With the 8x bit HIGH (5th bridge cut) and on a legacy mobo like KT266A or earlier, <b>then and only then do you run into the remapping function</b>. 7x becomes 15x, 7.5x -> 22.5x, 8x -> 16x, 8.5x ->16.5x, and all.

On something modern like nForce2 the setting 8x bit HIGH gives you access to high multipliers 13x and up, [<b>NO REMAPPING</b>. 8x bit LOW (5th L3 bridge closed) you just get the <b>LOW MULTIPLIERS</b>, 5x - 12.5x.

It depends on what motherboard you have.


<b>56K, slow and steady does not win the race on internet!</b>

 

[10GHZ]

i have a MSI K7T266Pro2-RU, which uses KT266A chipset. and yes, the highest multiplier that i can get is 15x. so i want to use the lower multiplier as higher ones.
so if my 5th bridges is already open, i dont need to do anything right? i can just use the 8x multiplier as 16x in bios??

 

[phsstpok]

Your bridges are fine the way they are. Don't close the 5th L3 bridge. All you will do is turn your 15x proc into a 7x proc and your BIOS options will be 5x - 12.5x.

Have you tried 9X multiplier in BIOS. This should give 17x just without any mods at all.

Lots of people are having trouble getting 16x and higher to work with FSB at 133 Mhz. Usually 16x works fine if FSB is 100 Mhz but of course you want the higher bus speed.

There are couple of things you can try. You can mod the proc to be a 16x proc (just for example) by changing the L3 bridges from L3 C-C-C-O-O to L3 C-O-C-O-O.

You can also try one of the workarounds (lower middle of page).

<A HREF="http://www.beachlink.com/candjac/Workarounds.htm" target="_new">http://www.beachlink.com/candjac/Workarounds.htm</A>

When I got my Tbred XP1700+ I mod'd the L3 bridges to make it look like an XP2400+. Fortunately it worked on my KT133A mobo but it only worked at the 15x to which I mod'd it. I later discovered the workarounds which I just mentioned.

These workarounds enable the illusive 13x, 13.5x, and 14x but also enable 133+ Mhz FSB operation with the other remapped multipliers.

I did a variation to KF's mod and connected a wire strand from AL25 (4x bit) to AM26 (Vcore). Now I have the 14x multiplier which I wanted and I can run my mobo upto 150 Mhz (the limit of my memory). This gets my XP1700+ to 2.1 Ghz.

I can also use 17x, and 18x but then I have to use a lower bus speed as my XP1700+ tops out.

I can't get 15x, 16x and 16.5x because of the particular connection I chose. Other mods should get you different multiplier combinations this is because forcing the multiplier bits one way or another change the expected remappings.



<b>56K, slow and steady does not win the race on internet!</b><P ID="edit"><FONT SIZE=-1><EM>Edited by phsstpok on 07/26/03 01:01 AM.</EM></FONT></P>

 

[10GHZ]

thanx for the feedback. just what i needed. ther was some problems trying to adjust the multiplier in bios, the bios has some bugs in it, it would hang the system everytime i try to restart the computer with x16 on. so i had to cut the L3 and now its running at 2600+ permanently and the system restarts noramlly. Seems like my Xp2400+ is running really solid with 16x multiplier. ran Sisandra Max3 Burnin on all of my CPU tests 10 times on HIGH priority and completed no problems. so i will try up the fsb later tonite.. thanx again

 

[phsstpok]

You're welcome and good luck with the OC.

<b>56K, slow and steady does not win the race on internet!</b>

 

[phsstpok]

By the way, the stability issues are related to the legacy multiplier circuits. The higher multipliers have timings intended for low bus speeds, as with the old Athlon 1.4 (14 x 100). These timings cause problems when bus speeds are high (133+ Mhz) <b>AND</b> multipliers are also high (13x +).

Somehow hardcoded multipliers and the workarounds fix this issue or at least avoid it.

<b>56K, slow and steady does not win the race on internet!</b>

 

[10GHZ]

hummmm.... looks like i learn somthing new everyday

got my 2400+ to run at 2200 yesterday 138x16, did the burn in on high 10 times with no problem. looks like i'll leave it there 4 now, till i get some better memory