BIOS edit, lost SSE instructions on AXP CPU

[caps_buster]

Long time ago, it was my first recap - a JetWay V266B mainboard, featuring KT266A chipset and Socket A compatible. The mobo recap went well, stability was regained and it can be even well overclocked.
However I did mod the bios a little - just changing the settings to what I usually set them, so bios reset was no issue and moving the default start date to present. That was it. However just recently I noticed the AXP 2800+ CPU (run at 146 x 12.5 ATM - so it show itself as AXP 2200+) with Barton core lack SSE:



And that is kinda weird. What could have caused it? Before it was NOT like this at all, there is the old screenshot:



...and as you can see, SSE is present.

The original bios I used to mod is there: http://ax2.old-cans.com/V266BA13.ZIP
The moded bios by latest ModBin & me is there: http://ax2.old-cans.com/V266BA14.ZIP

And the question is WHY?!

Oh, BTW, there is one more thing beside the bios mod. I was tried to force the CPU to x14 multiplier by hardwiring the CPU pins bellow the CPU socket. (using this guide: http://www.ocinside.de/html/workshop...md_pinmod.html ) It won't work, sadly, but could be affect the CPU instruction set features?

I used the CPU MSR to force enable SSE, and as soon as FFD show decoder stated to use it, the machine hard-crashed

Anyone who can tell me, what the hell went wrong?

[caps_buster]

LOL - the SSE instruction set was only disabled in BIOS settings, eh.
Feeling kinda embarrassed.



Anyway, I still have some questions to fullfill.

1) Can VIA KT266A chipset accept a x14 multiplier and work with it? (remember, last desktop multi is x12.5)

2) Can I update the microcode in the JetWay V266B bios to the latest version to fully support Mobile Athlons as well, as MP flag - Multiprocessing capable?

3) How can I check that the bios does not change multiplier settings to L6 settings before post?




http://www.cpuheat.wz.cz/html/AXP_mu...iplier_FAQ.htm
"In all cases, processor will start at L3 settings (startup multiplier). However, AMD wants notebook systems to load OS at maximum frequency, therefore BIOS on mobile systems is required to do software controlled transition to highest multiplier (that is to do P-State transition) immediately after initialization. On desktop systems, there can be two scenarios. The first one is that the BIOS does not do P-State transition. In that case processor will load OS at L3 settings. The second scenario is that BIOS program is somewhat "mobile-like" and it will transition CPU to maximum multiplier - that is to L6 settings. This this second scenario you can't just cut all L6 (that would give you 24x multiplier) but instead you have to cut L6 to a multiplier your CPU can handle at startup."

http://www.cpuheat.wz.cz/html/Photos...ier_Photos.htm
"Also, it is not good idea to open all L6 bridges at once because some motherboards have BIOSes that will change multiplier to L6 settings before POST."


To put things short, opening all L6 bridges on CPU give the maximum multiplier that can be set by software to 24 (eg. unlocing the CPU multiplier limits) and connecting L5 bridges give user Mobile Athlon and/or Athlon MP (Multiprocessor capable) ...

[caps_buster]

No reply? Well, I made some experiments and even some of them did not ended up very well, they does answer my questions.

So, the results are as follow:
1) VIA KT266A work with x14 multiplier just perfectly
2) JetWay V266B bios is maybe quirky and weird in many cases, but at least before post it did NOT set the multiplier to the L6 bridge settings on CPU - witch would with all cut L6 jumpers mean to attempt post at x24 multiplied FSB ...

I took a oldie Barton 2500+, unlocked bz default and it had breaked only one of the L6 bridge connections, so I cut the others as well and them - since I had already set the x14 multi by wires soldered on the bottom of CPU socket - it finally works and w/o any troubles. I finally saw a x14 multiplier on this old board. Unfortunately, this old suxxka made 2100MHz only with like 1.850Vcore and since the board did allow Vcore adjustments only for old Durons/Athlons (NOT XP ones), then I have to use only smaller FSB - but, here we go - 144 x 14 = 2044MHz!



I have to work a bit on the AXP 2800+, because there are the bridges kinda different from the old CPU's. The connections are much farer away and did seems to be somewhat submerged under the surface so... That is going to be hard.

[ibby]

Yes on the newer processors the connections are under a film!
be careful and keep thread updated.
you have not had a reply (becuase its old tech) but makes very good read!