those boards are all for the bin , save your time and don't bother with them....

This board must be sick and tired of me booting it up with the XP 2000+ CPU in it, that it's finally capitulated and given in, as it now boots up to the desktop without a problem. The only thing I did was to remove the default 0.0Volts jumper on the mobo, and stick it on the 0.1Volts jumper instead. I have no idea why I tried this - I just couldn't think of anything else to try.

I've briefly messed about with a few games, and all seems well. Speedfan tells me the CPU has reached a "nice n toasty" 50C, and rising. 3D Mark 99 Max tells me the CPU 3DMark score has risen to 26042.

I've got 2 other "bad" boards in the garage awaiting their fateful journey to the recycling centre. Perhaps it's time to rescue them, and mess about with that onboard voltage jumper, to see if shoving some extra volts through these faster CPUs can get these boards to work OK?

When the capacitors get warm they temporarily recover to some extend, so if you tried to start it a few times they may have reached that temperature. It will fail again when you let it cool down.

Do you know roughly how long it takes for caps to cool down? The mobo had been switched off for about 12 minutes, before I switched on the power again. Everything worked. Perhaps I should allow the board + components to cool down for at least 30 minutes?

BTW, 2 of the caps look "evil": bulging caps o' death. 😉

I left the mobo powered off for about 40 minutes, then powered it back on again. Everything seems to be OK, except for one thing. I just DL'd a benchmarking utility called Prime95. I ran it a couple of times, but it aborted very quickly, as it reported unexpected errors with it's mathmatical computations. So, I guess the board is unstable to a certain degree, but it seems capable of running games. So, all clear then! 😀

I just retested one of the other two "bad" boards - I selected the best looking one, the one that has had all its bad caps replaced. Before switching on the power, I removed the 0Volts mobo jumper, and put it on the 0.1Volts jumper. I tried the XP 2000+, XP 1700+, and also the 1400 tbird CPUs. All 3 froze the mobo within about 15 seconds. However, the 1200 (200fsb) tbird CPU seems OK. With this working 1200 CPU installed, I went to the "PC Health Status" area inside the BIOS setup area. The 5V reading seems low: 4.63V. Could this be the source of the problem?

h-a-l-9000 wrote:

When the capacitors get warm they temporarily recover to some extend, so if you tried to start it a few times they may have reached that temperature. It will fail again when you let it cool down.

That makes a lot of sense, it totally explains why my board can't take cold starts very well.

rg100, maybe just keep rebooting it till it stays. With mine i'll have to rebooting a coupld of times, and then it is stable as a rock. Although with a 5v line reading 4.6 isn't so good...

retro games 100 wrote:

I just DL'd a benchmarking utility called Prime95. I ran it a couple of times, but it aborted very quickly, as it reported unexpected errors with it's mathmatical computations.

Try bumping the I/O volatge (memory) to 3.45. This is one of the settings I need to keep mine going.

I've given up with the 2 "bad" boards - they work if you use a slow CPU such as a Duron or t-bird fsb200 CPU, but nothing faster, such as an XP CPU. So, these 2 boards are heading for the recycling centre. As for the one remaining "half working" board, well I powered it on again with an XP 2000+ CPU - it had been switched off for at least 12 hours. I left the onboard mobo volts jumper set to 0.1V (not it's default 0.0V), because that seemed to be a "lucky setting". Also, inside the BIOS setup area, I bumped up the IO voltage one notch, to 3.45. All seems well with it. Prime95 more or less instantly fails, but games work OK. Oh well. These Epox boards, they're good aren't they? 😉

It was defiantly an adventure. I had just stared researching them when I found one in a local shop. At this time I have it put away as some of those caps look like bombs. Maybe they always looked that way (since I got it), but I have other projects to deal with ATM. I can say that at it's final incarnation, I had it running for close to 2 months, as my main computer, playing games, and running Folding@home. So that means 100% cpu utilization, and it was at the 12.5x133 max with the memory running at fastest timings. With my 9088xt, I could play Farcry at 1280x1024, Medium settings /w 4xAA 8xAF @ 40 - 50 fps.

That's pretty good 😁

Far Cry is yet another famous game I've never tried. And so I make a quick pit-stop to Gamers Hell, and the demo's DL'd on to my trusty DVD-RW disc, and in to the retro rocket system. To quickly recap, my hardware is now -

nVidia GeForce 6800GT (I thought I'd give it another go!)

1 stick of 133mhz 256mb RAM (nothing fancy I'm afraid, so the mobo's BIOS memory settings are just left at the default "3 3 3" values)

XP 2000+ CPU

(BIOS "timings" for both RAM and CPU are set to 133.)

I was going to set up all the cool "AA" options etc inside that clumsy nVidia graphics control panel, but the Far Cry demo just won't run at all on my system. When I go to launch it, the game freezes. I suspect it's clashing with the 6800GT somehow. Nevermind. Instead, here's some final benchies for this beast before I too am going to move on to another project -

3DMark2001 SE (all its settings completely maxed out) = 8782
Quake 3 default demo (all its settings completely maxed out) = 126.9 fps
Quake 2 default demo (1024x768) = 252.2 fps

I just couldn't resist running another test! I managed to find a Thoroughbred core Athlon XP CPU. It's only a 1700+. I tried it on the board, and although it worked, the BIOS POST message reported it as being an "Unknow CPU". But I remembered that one of the 2 "bad" boards had a really unusual new BIOS datestamp of 04/09/2003. (The current one is dated 02/05/2002.) I saved this much newer BIOS .bin file using awdflash.exe, then flashed this current board with it. It worked! The BIOS POST now properly recognises the Thoroughbred CPU. I'm really excited about this. I might get another cheap Thoroughbred CPU on ebay. How about a 2200+ CPU? The other thing I'm excited about is that I'm going to try this t-bred 1700+ CPU in one of the 2 "bad" boards, to see if it will work. I'll do that ASAP...

By Jove, it worked! 🤣 I just tried one of the 2 "bad" boards again - the one that looks OK (with the new replacement caps), but would *not* boot up at all with a Palamino 1700+ CPU in it. With a Thoroughbred 1700+ CPU in it, I get no problems. And for the first time Prime95 doesn't instantly fail - it's running right ATM without any errors being reported.

I've gotta get myself some more of these T-breds. I wonder how high I can go with 'em?

Also: Speedfan says this t-bred 1700 is 63C. That's quite high. Perhaps it's because Prime95.exe was "cooking it"? Yes - I've stopped Prime95, and the CPU temp is rapidly decreasing...

Edit: I dug out that other second "bad" board. This is the one with distinctly rotten looking caps. I put a "simple" Duron in it, so as not to upset it too much, and proceeded to flash its BIOS with the "t-bred aware" BIOS. I then put the t-bred 1700 CPU in it, and it works, but only just. Prime95 fails every time, within about 20 seconds. SiSoftware Sandra sometimes locks up too. Games run OK though.

Edit 2: I seem to be having some success with the "rotten caps" board. Instead of increasing core volts in to this thing, I've done the opposite. Inside the BIOS, I *decreased* the core voltage by one notch setting. Now the board seems a bit more stable. I can run Prime95, and it doesn't fail instantly. The CPU also starts off cooler. I ran this test for several minutes, then got bored and stopped it. (But at least it didn't fail, although I realise this test needs to be run for ages.)

I guess the next step would be to get a low voltage CPU, such as a mobile t-bred. My guess is that it would help to stabalise a board which is unstable due to rotten caps. In any case, both "bad" boards have been spared their fateful journey to the recycle centre.

After reading your recent results, I'm excited for you. It makes sense about being able to lower the voltage, since the bad caps aren't being pushed as hard. I owuld ususally keep my Mobiel Barton running at around 1.475 volts, the max temp it would ever hit is 52c.

I guess the instabilities with these boards are indeed caused by the crappy capacitors. When my KTA3+ Pro was still new, it could pull off massive FSB overclocks (~170 MHz instead of 133) without missing a beat. Two years later it only managed 150 MHz, constantly degrading further, until it went completely dead with several caps bulging.
Near the end of the board's lifetime I also experienced the phenomenon that slightly lower voltages got me a more stable system.
Another problem with these old Athlon boards is that they generate the CPU voltage from the +5V PSU rail. Most ATX PSUs were designed for heavy loads on the +12V rail instead (ATX12V), it's rather difficult to find one that delivers stable +5V under load.

Meanwhile I managed to

• revive my KTA3+ Pro by exchanging the bulging capacitors with replacements from dead boards.
• kill the AGP port by severing a trace near one of the caps I was working on. 😵

Ah well. I'll keep this one for practicing my soldering skills. 😉
Replacing the caps is not very difficult, but you need good equipment and it's very tedious work. It took me more than two hours just to replace six caps.

I think I'll pay to have mine recapped/refurbished at some point, maybe find a one in better shape and have that done.

I was pleased (and a bit surprised) that a green coloured Athlon XP 2400+ T-Bred CPU (code AXDA2400DKV3C) worked inside another KT133A chipset based board, a QDI Kinetiz 7E-A. I then tried this CPU in the Epox mobo, and it was only then that I realised its BIOS only seems to support up to a 14x multiplier CPU. I checked this CPU on cpu-world.com, and it uses a 15x multiplier. I also found out from this website that the 2500+ CPU (AXDA2500DKV4C) uses a 14x multiplier, but there aren't any available on ebay ATM.

It's a pity that I have no XP-M (mobile) CPUs, either T-Breds or Bartons. I checked ebay, and XP-Ms in their T-Bred form are rare. Bartons are more common. Maybe I'll get one, at 2200 speed. I could always try it out in the QDI board as well.

At some point in the future, I will be able to test an Abit KT7A v 1.3 board. (I'll have to wait a bit for its arrival.) Then, that will complete one big happy KT133A family!

Finally, ATM I am testing the Epox board with the caps that look the most rotten. I notice that the mobo emits strange very high pitched "metallic whining". It's as if some of the failing onboard components are emitting their final "death throes".

Man, you have done almost the same thing I did in processor selection. Since the bios showed a 14x setting, I picked up an XP-m figuring I could change the multi as needed, and thought that if anything the default. Bottom line, If you using a mobile chip it won't boot at it's default, it will have to be set form the BIOS, or theorectily with a wire-mod.

The problem is that all the Via chipsets up until the KT400 only have a 4 bit control signal, which gives you a max of 12.5. The 14x setting in the BIOS is actually meant for an Athlon 1400+, which I read somewhere uses a different setting like 7x or something, and maps to 14x. Only those chips are going to work that way.

What I'm not sure about ATM, is if you get a higher than 12.5x non-mobile, if it will boot as a it'd default, and also if you get a mobile and wire-mod it to get the higher setting.

I just couldn't resist it - I've bought a couple of XP-M mobile Bartons:

AXMD2200FJQ4C (2200+) cpu-world says it's vcore is only 1.35
AXMH2400FQQ4C (2400+) vcore: 1.45

It doesn't matter if the 2400+ CPU fails in the Epox mobo, because I can try it in the QDI Kinetiz board instead. BTW, I have enjoyed reading everyone's posts on this subject.

Inside the BIOS set up area, I went to the Frequency/voltage control section. For the VCore voltage value, I selected the option that was "3 notches" less than the default setting, which is -0.075V. That gave me a Vcore setting of 1.425 for this particular BIOS value. (This is for the currently installed t-bred 1700+ CPU, code AXDA1700DLT3C, which cpu-world.com tells me is 1.5V vcore.)

Speedfan readings -

Core = 1.50V (spot on!)
+3.3V = 3.50V
+5V = 4.93V (not bad!)
+12V = 12.66V

The board seems to be stable. The only issue I have with it are the strange "metallic squeaking" sounds coming from where that large cluster of caps and metal things are located next to the ATX power cable socket + CPU. Perhaps if I had the caps replaced, that noise would stop?