I believe I’ve found a solution to a problem that is causing many headaches to some overclockers out there.
That is the dreaded KT133A issue, which is running a high-speed Athlon B at 133 FSB.
I bought a AVIA stepping 1.2 GHz Thunderbird, which is supposed to overclock quite well. I would have bought a C revision version, but since these cannot yet be obtained in the local market (Argentina), I chose to go with a Thunderbird B.
The only KT133A board available was the Epox 8KTA3/8KTA3+, so I bought that. However I was very dissapointed that I could not run my CPU at 133 FSB, since the motherboard would actually try to POST first at 12×133, no matter what multiplier was set in the BIOS. So I followed the recommendations at www.anandtech.com, and closed the remaining L6 and L7 bridges so the default multiplier would be 11 and the voltage 1.85v. This should help to make the POST easier (11×133=1466 at 1.85v instead of 12×133=1600 at 1.75v)
However this would only work 1 out of 20 boots. So I began my search on the web to find a solution.
A few days ago there was a post at www.overclockers.com saying that there was a rumour that this problem could be avoided if you just removed the 100/133 FSB jumper on the affected boards. The rumour was disregarded since apparently it did not work.
When I read that, I remembered that one of the BIOS revisions for my board did let me choose any FSB I wanted, without having to change the jumper. However this was useless at that time because anytime I tried to boot at anything above 120 FSB the system would not post.
However, if you use this particular BIOS revision, and you remove the jumper, then the board becomes jumperless. Very simple.
In my board’s case, the BIOS is version 30C12, and can be downloaded from www.epox.com. This is the initial production BIOS. It does not contain some minor fixes that the newer revisions have, and the voltage is limited to the +/- 0.050 range from the default voltage. This would mean you would not be able to go any higher than 1.800v, but if you close the remaining L7 bridge in an Athlon 1.2, then you can choose between 1.800, 1.825 and 1.850 volts. That is quite workable 🙂
In the case of other boards with this problem, hopefully there is some BIOS release with this capability. It does not mean that it will work the same as the EPOX board, but it’s something you could try. It’s worked for me just fine after two weeks searching for a solution other than cutting the copper bridges. Way too dangerous for me. I’m writing this in my Athlon 1.2 at 10×133.
PS: BTW, this is not an April Fool’s joke.
Editorial note: Well, if it is, they know where to find you.:)
Seriously, the originator of this did have this board, and probably did the same thing. His only mistake was assuming it worked all the time for all similiar boards. If you have an 8KTA3 and this problem, might as well try it and tell me whether or not this works.