Let me see this black goo.

I'd say it's only about 1mm long. My crappy camera would never be able to pick that out. It's absolutely jet black, and "shines" a little bit in torch light. I think I have cooked them both. Those "bumps" on the underside of the CPUs look kind of ominous. I'm wracking my brains to think what happened. Another theory is that the multiplier got "locked" on a low setting in the BIOS, eg 6x. Then, when the t-bred went back in to the mobo, this low multiplier didn't get increased. Instead, the CPU FSB got hugely OC'd in an attempt to get the thing running at top speed. I really need to figure out what did go wrong, because I want to try this out again. (But ATM I'm all out of 2400 t-breds! 🤣)

I have got a 2000+ and 2200+ t-breds. It's not funny smelling burning 1 second after powering on a mobo, so ATM I'm kind of reluctant to give these other 2 CPUs a go!

I've never smoked a CPU. I have a 1600+ that been running nonstop for years, but even with a Soyo Dragon board I never got the thing to overclock more then 100mhz. I tried and tried to overclock the thing even with crazy voltages but I never cooked it, it just wouldn't boot.

Yeah, you smoked them alright 😀) No worries mate - I've smoked a couple of them in exactly the same fashion as you did. Dont worry about the smell either. When you close the lid, you cant feel it that much 😜

...but it does linger 😁

If they failed I'm sure it had to do more with voltages then multipliers.

Lesson to be learned: Clear the CMOS when changing out CPUs.

I tried my 2000+ and 2200+ t-breds (in the abit kt7a 1.3 non raid board). The 2000+ CPU POSTed OK. The BIOS CPU "default" settings tell me its running at "2000+ rated" speed @ 133fsb. That's correct. I tried the 2200+ CPU, but I got no POST. Regarding the 2400+ t-breds, I wonder if there is something a bit "special" about them? I wonder if I don't have to do any "magic tweaking" in the BIOS, but simply to buy another one, put it in, and switch the power on? - and it will work OK. I wonder if there's just something "unusual and lucky" about the way I've had good success with the 2400+ t-breds, and that attempting to get a 2200+ t-bred to work might always fail? - at least, inside these slightly odd and fussy Abit boards.

I think I'll stop messing about trying to get a 2200+ to work in this board, and just buy another 2400+ t-bred. Luckily, there's plenty left on ebay. When it arrives, I will simply clear the CMOS, put it in, and switch on power. I'm hoping that it will "just work", like I had it working earlier yesterday, before my additional (and unnecessary) experiments became fatal to both of these 2400+ CPUs.

I got a "new" t-bred 2400+ CPU, to replace the ones I accidentally cooked last week. It works OK in this Abit 1.3 board, and runs at 133FSB with a real clock speed of 2000Mhz. Here is a step by step list of actions that resulted in the successful installation of this CPU -

1) Removed then replaced the button battery.
2) Cleared CMOS jumper. Hehe, I'm not taking any chances.
3) Install the 2400 CPU, then switch on power.
4) BIOS POST should appear on screen. Screen says CPU is XP 1800+. Don't worry! (Interesting side note: Tried a 2200 CPU lots of times. Never once saw a POST. The 2200 and 2400 have the same core voltage of 1.65. Strange!)
5) Hit delete key, to access the BIOS set up area.
6) Choose "fail safe", and also set that "quick boot" thing to disabled, so that you can see the RAM counting test on the screen. (This just gives you a bit of time to press the delete key again, which needs to be done in a moment.)
7) Save and exit BIOS.
8 ) BIOS POST should appear on screen. Press delete key again to access the BIOS set up area for the second time.
9) Do a few very *basic* changes, eg disable the floppy disk. Nothing fancy. No messing with voltages, CPU speed, anything like that.
10) Save and exit.
11) BIOS POST should appear on screen, and the CPU is now correctly reported as an XP 2400+, running at 133FSB with a real clock speed of 2000Mhz.

Basically, you need to switch on the power, and "mess about" in the BIOS twice in order for the mobo to get it's act together to find the XP 2400+ correctly. You don't need to mess with the important CPU related BIOS settings. It will do all of this automatically, just not immediately.

There's something unstable about this KT7A 1.3 non-raid board. I've run 3DMark2001 second edition several times, and twice it's dropped back to the desktop. I've never seen that happen before. (Also, I can't install an ATI Radeon 7500 in it, because the win98 desktop gets corrupted.)

I'm using the VIA Hyperion 4-in-1 4.56 driver package. The current graphics card is an nVidia MX 440 AGP card. I've run 3DMark2001 S.E. once more, and this time it completes. Its score is 4888.

retro games 100 wrote:

There's something unstable about this KT7A 1.3 non-raid board. I've run 3DMark2001 second edition several times, and twice it's dropped back to the desktop. I've never seen that happen before. (Also, I can't install an ATI Radeon 7500 in it, because the win98 desktop gets corrupted.)

I'm using the VIA Hyperion 4-in-1 4.56 driver package. The current graphics card is an nVidia MX 440 AGP card. I've run 3DMark2001 S.E. once more, and this time it completes. Its score is 4888.

Have the same issue with my v 1.1 board + T-bird 1000....it crashes to the desktop with the same gfx card you're using and with the GF FX 5950 Ultra
I'm using the same VIA driver version but I d/l mine from Abit's FTP server

Hmmm, that's very interesting! I wonder what's going on? Let's see if we can solve this problem. Some questions -

What FSB are you running at? (I am running at 133)

What RAM are you using? (I am using 1 stick of PC-133, 256mb)

What BIOS settings are you using? (I am using "fail safe", but I also tried to "boost" my AGP settings by setting some of them to Enabled. Eg: AGP 4x mode is Enabled.)

I gave Prime95 a quick go, but got no errors. Usually, when a board is unstable - eg an Epox mobo with capacitors that look like bombs, Prime95 fails quite quickly. I wonder if this problem is just related to the AGP?

retro games 100 wrote:

Hmmm, that's very interesting! I wonder what's going on? Let's see if we can solve this problem. Some questions - […]
Show full quote

Hmmm, that's very interesting! I wonder what's going on? Let's see if we can solve this problem. Some questions -

What FSB are you running at? (I am running at 133)

What RAM are you using? (I am using 1 stick of PC-133, 256mb)

What BIOS settings are you using? (I am using "fail safe", but I also tried to "boost" my AGP settings by setting some of them to Enabled. Eg: AGP 4x mode is Enabled.)

I gave Prime95 a quick go, but got no errors. Usually, when a board is unstable - eg an Epox mobo with capacitors that look like bombs, Prime95 fails quite quickly. I wonder if this problem is just related to the AGP?

Using 100mhz FSB since it's a 200mhz CPU

Using two sticks of 512mb 133mhz RAM

Have tweaked most of the BIOS settings using this guide:
http://www.devhardware.com/c/a/Motherboards/K … OS-Tweak-Guide/
Used these settings on my last, now dead KT7A mobo
So you can say I "maxed" everything in the BIOS...even the AGP stuff

I've ran prime95 too without any crashes or lockups and my caps looks like when the board left the factory

OK, I reckon it's time to roll back these VIA chipset drivers, to an earlier version. I'm going to overwrite my 4.56 package with an older version. First of all, I'm going to try 4.35. I'll install it, then retest 3DMark2001 S.E. and see if it crashes back to the desktop...

Edit: Nope, that didn't work. I uninstalled the 4.56 package, then installed the 4.35 package. I then ran 3DMark2001 S.E., and it never reached the end of its tests - it just dropped back to the desktop.

I tried 2 other things, both of which sound like they are "clutching at straws", but may have some merit in solving this problem.

1) For the CD-ROM drive, I unticked the "auto insert notification" option. This can be found inside Control Panel -> System -> CD-ROM -> Settings tab

2) Inside Control Panel -> Power Options, I switched off every power option.

I then reran 3DMark2001 s.e. 3 times in a row, and all 3 tests worked. However, I noticed that some of the animation was not very fluid: sometimes the animation would "stick" for a fraction of a second, other times it would momentarily speed up. Also, my scores were quite different each time: 41??, 42??, and 4828 (?? means I can't remember the value, but it doesn't seem important. Also, it's strange how that 3rd and final test appeared to "enjoy itself" more than the first 2 tests.) The testing utility didn't drop back to the desktop at any stage.

Problem solved perhaps? It could either be something weird going on with the CD-ROM, or a power saving problem, or both, ...or something else! (The reason why I wonder if it's to do with the CD-ROM is that sometimes it seems to spin up or down at "inappropriate"/illogical moments.)

One thing of interest: inside the Control panel -> System -> System devices area, I cannot see a system device with the word "Gart" in it. I can see a system device called VIA CPU to AGP Controller. Is this all I need?

Edit: I just ran 3DMark 2001 s.e. for the 4th time in a row, and it worked OK. In fact, I got 4828, which is identical to the score in the 3rd test.

I switched off the machine for a few minutes, then back on again. I just wanted to rerun 3DMark2001 s.e. a couple more times, to see if the problem had gone. I ran it twice more. Test result 1 score was 4090. The problem with this score is that it is inconsistent with all of the other scores. Something is not quite right. They shouldn't all differ like this. Test result 2 = the test utility went mad! On the test where there is a spaceship in the sky, and it gets shot at then falls to the ground. That falling animation was endlessly replayed. It had got stuck. I had to reset the machine. Either it was a mad bug inside 3DMark2001 s.e., or the mobo/AGP had gone mad. I think VIA should have written an "anti-mad" patch for their driver, in an attempt to "un-insane" it. Or perhaps it was too difficult for their programmers?

Edit: Oops, very sorry. I may have got my first edition and second edition testing muddled up. When I was referring to the s.e. tests above, I think I was running the f.e. (Both desktop icons look the same.)

I just tried the s.e., and the animation appears smoother. I get a score of 4503, which sits right in the middle of all of my tests anyway, so no harm done. BTW, 2 more tests worked - this test, and another previous test using the f.e.

Edit 2: Just reran s.e. (yes s.e., not f.e.), and I get 4548, which is quite close to the first s.e. score of 4503. And no problems running test.

Edit 3: I reran the s.e. once more, and it crapped out with a typical windows 98 "illegal operation" error message pop up box.

Edit 4: After the above "illegal operation" problem, I had to reset the machine because it had frozen. I then ran 4 different versions of 3DMark in a row: 2000, 99, 2001 f.e., then finally 2001 s.e., and they all worked. I say "worked", they didn't crash or drop back to the desktop.

I think the board is more stable now. I don't know if it is because I'm using driver version 4.35 (not 4.56), or whether those other tweaks have helped a lot: that is, disabling the CD-ROM "auto notification" (see above), or switching off all of the Power Options (again, see above).

I'm waiting for the VIA chipset haters to say something... 😉

I'm bored of VIA-bashing, but you guys are definitely silly for banging your head on buggy VIA stuff.

HunterZ wrote:

I'm bored of VIA-bashing, but you guys are definitely silly for banging your head on buggy VIA stuff.

Yeah but, 97589 bungholioMarks don't lie!

That Soyo Dragon board I ran with a 1600+ used a VIA KT266a, I never had issues with it. So has VIA gotten better or worse? I just bought a VIA board off newegg for $15 +free shipping after promo code. http://www.newegg.com/Product/Product.aspx?It … -L8A-_-13138170 They just sold out though.
For 15 bucks what I got too loose? 😁

bushwack wrote:

So has VIA gotten better or worse?

I believe they got better. 😀