Here goes the BIOS.

Hi all,

I've been going by this thread to get my BX2000+ running reliably with a Tualatin Celeron 1.2 CPU, and am encountering similar problems as the OP. Even with the "SpecialBios" as above, I can't get it running reliably at anything else than 100FSB. It certainly posts @ 112, 124, 133 and every now and then it'll get past the memory check and will boot Windows and be perfectly reliable. Then I'll try again and it'll freeze at the memory check, come up with a DualBIOS error, or maybe get to Windows boot but hang with a GPF error.

The ASUS slotket I'm using is untouched, set to 1.65V, only the CPU has pins removed for Tualatin support, I have not done the AK4 --> AK26 wire mod yet. Like others have noted, it seems there is issue with some timing of the fast CPU on this board.

Any other tips?

Thanks

JB

first of all, did anyone check the original website of this adapter?

https://web.archive.org/web/20061217061510/ht … mpatibility.htm

GA-BX2000 v1.1 Must use f9 BIOS.
GA-BX2000+ v1.2 Must use fdc BIOS.

the fdc bios file can be downloaded from archive.org, and its contents are different from the special bios posted on this topic.

did anyone try the correct bios?

Batyra wrote:

Here goes the BIOS.

I'm in the same situation you where in, but my CPU is an Pentium 3 1000 MHz 133FSB. I wonder if I flash the special bios from you will solve the problem... What do you think?

Just to add my two cents. Same problem here with Gigabyte GA-6BXE (I'm starting to think that's the issue with Gigabyte boards, just as the Korean guy told someone here earlier in the PM). I'm using that MSI slotket MS69something v.2 and a P3S-1400, that I bought from the Korean guy. It posts fine on any FSB including 133, there are no problems like on BX2000. However I did patched its bios with a bios pather from that Belarussian forum. It added proper microcodes and detection routines. It does supply proper 1.45 Vcore without any need to set jumpers.

However! It's unstable at 133FSB. And that pertains only to that specific P3S (both SL6BY and the earlier SL5something can't remember now stepping). Cumine P3 1000mhz socket 370/133 works fine and rock stable on this board in the same slotket on FSB set to 133mhz. P3S is not 🙁 I've tried different video cards (namely GF2, GF4, GF4ti, Voodoo 3, Voodoo 5) It posts 100% of the time. Windows 98 boots fine. Any desktop activity and even aida64 test runs alright. Until you get into full screen DOS gaming (e.g. Carmageddon) or run Prime95. It locks up in about 1-5 minutes.

Ah well... 1050mhz at 100FSB it is, I guess 🙁

Zoomer wrote on 2020-01-17, 21:10:

Just to add my two cents. Same problem here with Gigabyte GA-6BXE (I'm starting to think that's the issue with Gigabyte boards, just as the Korean guy told someone here earlier in the PM). I'm using that MSI slotket MS69something v.2 and a P3S-1400, that I bought from the Korean guy. It posts fine on any FSB including 133, there are no problems like on BX2000. However I did patched its bios with a bios pather from that Belarussian forum. It added proper microcodes and detection routines. It does supply proper 1.45 Vcore without any need to set jumpers.

However! It's unstable at 133FSB. And that pertains only to that specific P3S (both SL6BY and the earlier SL5something can't remember now stepping). Cumine P3 1000mhz socket 370/133 works fine and rock stable on this board in the same slotket on FSB set to 133mhz. P3S is not 🙁 I've tried different video cards (namely GF2, GF4, GF4ti, Voodoo 3, Voodoo 5) It posts 100% of the time. Windows 98 boots fine. Any desktop activity and even aida64 test runs alright. Until you get into full screen DOS gaming (e.g. Carmageddon) or run Prime95. It locks up in about 1-5 minutes.

Ah well... 1050mhz at 100FSB it is, I guess 🙁

thats not the slocket used in the first post of this topic
the "upgradeware" one

if you had one of those you could try

https://web.archive.org/web/20080124114212/ht … mpatibility.htm

GA-6BXE v2.1 :The original BIOS does not work with a Tualatin processor. You must use GA-BX2000 f9 BIOS if you like to use a Tualatin processo

flash a bios from another board ? seems crazy

That's some interesting info, thanks! Definitely sounds dangerous though. I believe, most it will do is give me proper Tualatin detection - which I already have.

Still... Might try it 😀

i just got one of these boards and I'm working on getting it setup.

i started off with a 700 mhz slot 1 100fsb. with whatever bios version it came with stock it was complaining about health check status even though it was running at 1.65v it would post and boot though

so i updated both the main and back up bios to fdc bios posted above from archive.org its stopped complaining about the health check and changed to saying ok.

so then i swapped out the 700mhz cpu to a Tualatin Celeron 1.4 in an asus slocket it picked it up right away shows it running at 1.4ghz and 1.5v.

do i need to mess with any of the jumper setting on the block for setting ratios and fsb there is obviously no setting on there for a 1.4ghz chip does it even matter what that is set to?

second i have 2 256MB PC 100 cl2 ECC ram installed i manually had to change the cl from 3 to 2 in the bios. i turned on ecc mode i think. but nothing on the main post screen or that secondary one right before the windows splash comes up says anything about ecc being on. also in memtest86 v4.3.7 it also shows nothing about ecc in there

were can i make sure that ecc is working the gigabyte website says it supports ecc. on an asus p3b-f on that secondary post screen it shows there when ecc mode is on idk if gigabyte just don't do the same or if i'm not setting something right in the bios.

is there any default bios settings for this board that should be changed to get it running optimally.

this mb need ecc sdram for 133mh tulatin

Could you elaborate? Your thoughts could be helpful

Yes! I would like to bear more about this theory(?) more. @956229

Because every owner of a 440BX motherboard with 4 RAM slots has a strong desire to get 1 GB of RAM at a frequency of 133 MHz.
This can be achieved by sorting through a huge number of standard 256 MB RAM doublesided modules (preferably CL2*) or immediately selecting a ‘server’ modules.
However, the memory required is not ECC, but FB-DIMM (fully buffered, or ‘server’) with 18 memory chips (8+1 * on both sides).

*- Let us remind you once again that the 440BX does not have a memory frequency of 133 MHz, but rather a heavily overclocked memory of 100 MHz.
This means that at a frequency of 133 MHz, non-standard, significantly reduced operating times are used for RAM sub-timings.
FB-Dimm RAM handles this mode more easily.

I will not start a new thread because this one is already rich in the necessary information. This thread relates to tualatin but I will be considering coppermine 1000/133.
At least a few people here have reported that the board has trouble booting (except for Tualatin), Coppermine 1000/133, and the symptoms include a hang before the memory test. I don't know how it works with Tualatin, but with Coppermine, a partial solution turns out to be... wait.
This board ALWAYS works with Coppermine 1000/133, regardless of the BIOS version. If there's a hang before the memory test, you'll need to wait a while. Sometimes the board will boot immediately, sometimes you have to wait over two minutes. The average time between the BIOS screen and the memory test so far is 55 seconds. For me, it's not a major problem, but I'm curious what causes this behavior and how it could be resolved. Perhaps a "Korea mod"?
I ran tests on two BX2000+ boards, all of them is recapped by Panasonic FR. Both boards run 1000/133 rock stable, passing 3D Mark, HL2, Quake 3, and Prime95 for hours. Hang at startup is the only issue.
I haven't figured out yet whether different BIOS versions matter here, but it is visible that on the latest BIOSes the hang time is longer than on the older ones.
Request to Tualatin owners - If you experience a hang on boot issue, just wait and let us know if the system "moves on"

shevalier wrote on 2025-11-23, 07:30:

Because every owner of a 440BX motherboard with 4 RAM slots has a strong desire to get 1 GB of RAM at a frequency of 133 MHz. Th […]
Show full quote

Because every owner of a 440BX motherboard with 4 RAM slots has a strong desire to get 1 GB of RAM at a frequency of 133 MHz.
This can be achieved by sorting through a huge number of standard 256 MB RAM doublesided modules (preferably CL2*) or immediately selecting a ‘server’ modules.
However, the memory required is not ECC, but FB-DIMM (fully buffered, or ‘server’) with 18 memory chips (8+1 * on both sides).

*- Let us remind you once again that the 440BX does not have a memory frequency of 133 MHz, but rather a heavily overclocked memory of 100 MHz.
This means that at a frequency of 133 MHz, non-standard, significantly reduced operating times are used for RAM sub-timings.
FB-Dimm RAM handles this mode more easily.

I agree. This doesn't change the fact that the behaviors described here (as well as those I described in the thread about running s.370 coppermine on this board), have nothing to do with memory quality. Both of my boards boot with all PC133 memory, and P3B-F and BE6-II users also don't complain about having to select sticks.
And I would really like to check now if using Infineon 133/C2 will change anything in this matter, but unfortunately it was damaged 😒

dulu wrote on 2025-11-23, 17:52:

I agree. This doesn't change the fact that the behaviors described here (as well as those I described in the thread about running s.370 coppermine on this board), have nothing to do with memory quality. Both of my boards boot with all PC133 memory, and P3B-F and BE6-II users also don't complain about having to select sticks.
And I would really like to check now if using Infineon 133/C2 will change anything in this matter, but unfortunately it was damaged 😒

I have a CUBX-L and I am complaining. 🙁
During the period when socket370 was relevant, there were constant complaints on forums about 4 memory sticks.
From my experience
- Infineon/Micron CL2 - the result is almost guaranteed
- Samsung/Hynix CL2 - 50/50
- The rest or CL3 - a silicon lottery.
Therefore, if you are starting a collection from scratch, it is better and easier to get a set of FB-DIMM PC133. It will be cheaper and less stressful.
And the result is immediate and guaranteed.

PS
There is no mention of 133 MHz memory in the 440BX chipset settings.
It seems that “some setting related to RAM #1” is 0/1
“some setting related to RAM #2” is 0/1
All settings there occupy 1 bit — “0-66 MHz/1-100 MHz”.

dulu wrote on 2025-11-23, 17:52:

This board ALWAYS works with Coppermine 1000/133, regardless of the BIOS version. If there's a hang before the memory test, you'll need to wait a while. Sometimes the board will boot immediately, sometimes you have to wait over two minutes.

Perhaps only the reference Intel works perfectly with SPD of RAM sticks.
The same CUBX reports that it has detected SPD and (for some reason?) begins to calculate the memory capacity on its own.
As far as I remember, if POST is enabled, it calculates the memory three times, and if it is disabled, it calculates it once.
If the data in the SPD of the first module is not found, then the SPD is not used at all.
You can try removing the SPD microchip from the some unnecessary 64 MB RAM module and observe the motherboard's start-up time.