I removed the potentially defective T-Bred 2400+ rated CPU, and installed a T-Bred 2000+ CPU. Now when I run SiSoftware Sandra 2002 Pro, the Multimedia test works correctly. I get a typical score, and both fractal images are displayed correctly.

retro games 100 wrote:

Equipment. Epox EP-8KTA3+Pro mobo T-Bred 2400+ rated CPU Voodoo3. Also tried a GF4 SiSoftware Sandra 2002 Pro […]
Show full quote

Equipment.
Epox EP-8KTA3+Pro mobo
T-Bred 2400+ rated CPU
Voodoo3. Also tried a GF4
SiSoftware Sandra 2002 Pro

Result.
On the Multimedia test, I get:

Integer aEMMX/aSSE 975 it/s
Floating-Point aSSE 10684 it/s

The Integer result is approximately 10 times less than it should be. Also, the floating-point result should be a bit higher. When the two fractal images are displayed, one of them has corrupted graphics. I wonder if my T-Bred CPU is slightly damaged? One of its side edges looks very slightly "crumbled". BTW, I should be getting on with other work, but testing retro hardware is more fun than doing real life chores!

Edit: I've just noticed something else. It's not related to the problem above, but doesn't really warrant a new Vogons thread. This Epox board won't POST with a low powered 1.35v XP-M CPU. It's a 2200+ rated CPU. The interesting thing is that it POSTs fine on another identical board. The difference between the two boards must be the quality status of its capacitors. So, if you can't POST with a low powered CPU, it might work if the caps are fixed.

Both boards have the exact same revision and bios version also?
And about the defective tbred, usually a slightly crumbled core doesn't have to mean it's defective. The real core is a little bit smaller then the die itself, a 'buffer', so to speak.
I'm not sure about the bad caps, might it be that the mobile isn't officially supported?

I know this 'can' be an issue, for example:
The Asus A7V133 1.05 (without the dot, theres also the "A7V133 1.05.") doesn't officially support Palomino but I've read that some will work with them and some won't.

Another example:The Asus P2B 1.12 officially supports Coppermine and the 1.10 doesn't. But there are 1.10's with the updated PLL chip (I think it's that one) which in fact will run Coppermine. Otherwise the 2 different 1.10's are identical.

Maybe you got 2 boards which should be identical, but aren't 😉

I must be honest here. The problem may be "pilot error". It's true that both boards are the same. But it's possible I may have put in a 2200+ T-Bred, and not my 2200+ mobile barton. They look very similar - both a kind of dull brown colour. I now think the 2200+ T-Bred is completely broken, and this is why it won't POST. I might try the 2200+ mobile barton later.

However, regarding the 2400+ T-Bred situation: I do think it's defective. I removed it, and put in a 2000+ T-Bred, and that instantly fixed the Sandra Multimedia test. There's no doubt about this. 😀

Sorry for any confusion. 😦

Update: I just want to confirm that the xp-m 2200+ CPU does work OK in the mobo. I accidentally put a broken t-bred 2200+ in to the board, and wondered why it did not POST. I got these two CPUs mixed up, because they look very similar in appearance. Sorry for any confusion.

That's the cool thing about Retro computing...

Busted cpu? Who cares! I got another one lying around 😀

Tetrium wrote:

And about the defective tbred, usually a slightly crumbled core doesn't have to mean it's defective. The real core is a little bit smaller then the die itself, a 'buffer', so to speak.

At the moment, I'm testing a socket 370 Pentium 3 processor. Its ID code is SL5QV, and it has an "exposed" die, because it does not have a built in heatspreader. Before applying thermal paste to it, I cleaned the die and its surrounding area. After cleaning, I noticed that one of its four sides was "bashed about" looking. In fact, this edge looked in rather poor shape. Nevertheless, the PC POSTs OK, and I am currently running "Prime 95", and no errors are being reported. I guess this "bashed about" side could have occured because of a carelessly mounted heatsink. Also, I'm hoping that the CPU itself is not damaged, because the area affected only appears to be on one of its edges.