elfuego wrote:

If you are willing to pay 5e + postage fees (from Germany) and wait until end of holidays (15th of January), I can send it to you. For 1e more I will include the 2400+ 😀

Thanks for your kind offer! 😀
Yes, I'm interested. Will contact you via PM tomorrow.

retro games 100 wrote:

Christmas day sounds fun at chez prophase_j's! 😉 Now that the dust has settled 😉 , do you know what happened? I wonder if you fried the CPU, because the mobo was anticipating a previously installed CPU?

I have no idea. The board hadn't been used for long time, and technically hasn't been at anytime of my possession. I had just installed a new CMOS and cleared the jumper, as is my standard operating rules my first time with any motherboard. In addition to the power supply and processor, I also had my Voodoo 5, MX300, and a random 512 stick of ram.

Pretty much everything was assembled in the case. At this point I hadn't installed a hard drive, the only connections made were KVM, power coard and switch. And as soon as I pressed the switch...

Based on my observations I expected the failure to have came from somewhere in the middle, but I really can't pin it down. My best guess places them in the top right corner, where the domes on two capacitors are literally split open, something I can't recall seeing before.

As far as the processor and other peripherals everything is still working, accept the memory stick, only because it hasn't been tested.

gravitone wrote:

The kt133(A) is for all intents and purposes, the most unholy chipset of them all.

The chipset IS unholy only when you cannot play with it 😀...I mean if the bios is a FAIL, no tweaks for memory subsystem, no tweaks with the chispet/memory timings, no OCing at all...that is a BS...but take a fine example Abit VP6 and take a look into the BIOS - you'll get what the KT133 needs to perform as well as any intel chipset...Of course via was always few steps behind with memory bandwidth, but this can be tuned and may perform as well as intel...So I guess it's just about good board and components, not the chipset itself. I beat a lot of DDR setups with my SDR PC166 😀

Shodan486 wrote:

I beat a lot of DDR setups with my SDR PC166 😀

Probably due to the increased bus speed. 😀

swaaye wrote:

Shodan486 wrote:

I beat a lot of DDR setups with my SDR PC166 😀

Probably due to the increased bus speed. 😀

That's why I pointed out the the chipset and (IF available) good bios combo...and this is the result. I feel resentment everywhere when the word VIA is spoken & especially the older ones, mainly the KT133 - people just can't understand that while Intel has all-compatible-but-almost-unscalable chipsets, VIA's chips only perform well enough only if they can be adjusted to our will.

I'm not sure I'd defend VIA's hardware, regardless of the fact that lots and lots of tweaking and BIOS workarounds can make it work reliably and fast sometimes. 😁 I also don't agree with saying Intel chipsets aren't tweakable because 440BX and 815 do respond very well to it. 815 in particular. Too bad about the lame 512MB limit, but it wasn't a big deal in 2000/2001.

If I am going to mess with Socket A, the only chipset I really want is nforce2. Overclocking without cranking the AGP and PCI to the sky and exceptional memory performance. There aren't any issues really. This chipset also just feels snappier than any other Socket A chipset that I've used. Even compared to KT880.

I've blocked out the "good old days" of VIA for the most part. I do have a VIA K8T800M notebook and maintain a K8T800 desktop. Those are decent chipsets, probably because AMD took the hard stuff out of their hands. HDD I/O has always seemed a little slow on the VIA southbridges however.

Yes, the only exeption at that time was the 440BX (of course I am aware of the older sister of my config, the Abit BP6), the way it could be cranked up so high really startled me when realizig it were the Celerons that ran on it (celerons with the cache, not their L2less older brothers, were basically at the same level of overclockability to my experiences), and you're right, the 815 was a quite fail with that mem. limit.

Nforce2 RULEZ! There is no objection to this marvelous miracle of the AthlonXP ages. I really loved it, though never played with it. Instead, I have a lot experiences with the shitty KT333 and KT600, both unable to run at 400FSB (only with the RAMs from the recommendation list), anyways it was a low competitor against the nVidia's solution.

KT800 and 880 are a very good chips, will run stable at 800fsb no matter what. To conclude you, I absolutely agree with the slow southbridge performance, but this can be of course fixed using own PCI SATA add-on card, no? ( nowadays ) 😅

VIA wasn't that bad in the KT133 era, it provided a cheap solution, but from the technological point of view, it is a white knight - it came as a cheap and sufficient solution (and just in time), supporting both Intel and AMD CPUs instead of expensive chipsets of these two gigants. Applause !!!

"at the same level of overclockability level as the PIII were" -just to add...

Shodan486 wrote:

Nforce2 RULEZ!

Word. One of the best mobos I've ever had, and one with which I had most fun was Abit NF7S V2.0. Paired with a few good sticks of DDR 400 and a mobility Athlon XP it offered incredible performance. Even that onboard soundstorm was impressive. I've trashed my Live! because I liked the sound of SS better.

I have an XP-M 2200+ CPU. I also have the CPUMSR software running under Win98. I can increase the CPU's multiplier to 12.5x, using CPUMSR. But which level 6 bridge(s) do I need to cut on this 2200+ (1.35v lower powered) CPU, in order to get access to higher multipliers? Looking at the table entitled "Reference, L6 : SFID for Mobile" on this webpage -

http://fab51.com/cpu/barton/athlon-e24.html (about 3/4 the way down)

, I'm not sure which L6 bridge(s) need to be opened, or closed?! Thanks a lot for any advice.

"When L5 [1] bridge is closed, L6 bridge becomes invalid and the maximum multiplier is set to 24X."

This means you can get access to all the multi's, and as long as you don't try to use it in AMI bios board you should be okay.

Thanks a lot. I now need to get some "electric conductive insulator", in order to close L5 [1]. (Do I also need to close L5 [2], in order to set "mobile mode" to "true"?) I tried looking on ebay for electric conductive insulator (and similar phrases), but my searches showed no results. Any ideas please? Thanks!

BTW, I'm testing an Epox EP-8KTA3+PRO board. If I wasn't busy with a shelving project (guess what for, hehe!), I'd have a go at my first caps replacement on this board. They are in a shocking condition. With a Palomino 2000+, I never see the win98 desktop because it reboots each time. If I set the core voltage to -0.1V (it's lowest setting; very useful to be able to set the core voltage to "minus something"), then the XP-M (1.35V) CPU is stable. It runs at about 1.51V. It boots up at 6x + 133FSB = 800Mhz, and using CPUMSR, I can set the multi to 12.5x. BTW, my Sandra 2002 Pro memory bandwidth scores aren't too shabby. That's using CL2 sticks. (Sorry, how does "Print Screen" work? I pressed the "P.S." button on the keyboard, but nothing happened.)

BTW, elianda, if you are reading this - I set up this Epox mobo yesterday with a "basic BIOS configuration", where I did not "max out" its memory timings. I also did not crank up the multiplier on the XP-M CPU. I had no sound delays testing the EWS. Now, I have "maxed out" the memory timings in the BIOS, and I also set the XP-M's multi to 12.5x, and I am getting a few odd sound delays with my sound testing. I will continue to do more testing with the BIOS memory settings, and the XP-M's multiplier settings, in conjunction with the EWS sound tests...

I have done some more testing with the EWS sound card, in conjunction with altering the XP-M CPU's multiplier setting. I can consistently get these events to occur:

I do not set the CPU's multi using CPUMSR. Therefore, the CPU's multi is left alone, at 6x. dxdiag sound testing works - no delays.

I set the CPU's multiplier using CPUMSR. I set it to 12.5x. dxdiag sound testing always produces delays.

I set the CPU's multipler using CPUMSR, and set it back to 6x. dxdiag sound testing is "magically fixed" - no sound delays.

Edit: I use CPUMSR, and set the multiplier to 10x (1333Mhz). Sound appears to unaffected by delays - no problems. If I set the multiplier to a value greater than 10x (eg 11x), sound delays begin to "appear". If I set the multiplier value to maximum (12.5x), the sound delays are significant, and occur regularly.

This could be an "ISA voltage issue", as elianda suggested. Perhaps when you increase the CPU's multiplier, the mobo's "voltage resources" could be "consumed" by the CPU's increased requirements, and not enough stable juice diverted to the ISA slot? Sorry for this uneducated sounding description!

prophase_j, did you once report on Vogons that you could not get sound to be heard from your AWE64 ISA card, when it was installed inside your Epox 8KTA3 KT133A based mobo, using an XP-M CPU? If yes, then I wonder if I am experiencing the same kind of problem. I am testing an EWS ISA sound card, inside an 8KTA3+Pro mobo, and it will not function correctly if I set the CPU's speed to a fast value, by increasing its multiplier using the CPUMSR.exe program. If I lower the speed of the CPU (by reducing the multiplier value), the sound card behaves itself.

Edit: This is interesting - the "cut off" point seems to be a multiplier of 10.5x. This gives a real clock speed of exactly 1400mhz. The sound card behaves itself OK at this point. Anything beyond this point results in failure = sound delay problems. So, if I increase the multiplier by just "one notch" from 10.5x to 11x (which gives a real clock speed of 1467mhz), then the sound is messed up = sound delay problems. If I then reduce the multiplier from 11x back to 10.5x, then sound is "magically fixed", and I get no sound delay problems.

Does the ISA socket "go a bit mad" if the host CPU is operating at a speed of beyond 1.4Ghz?!

Maybe it's just buggy drivers that can't keep up with the CPU speed.

What happens if you reduce the FSB to 100MHz and use a high multiplier (e.g.: 100x12.5)?

5u3 wrote:

Maybe it's just buggy drivers that can't keep up with the CPU speed.

What happens if you reduce the FSB to 100MHz and use a high multiplier (e.g.: 100x12.5)?

Unfortunately, I can't reduce the FSB on this Epox mobo! If you put in a "fast type of CPU" such as an XP-M, then the FSB is set to 133, and there is no option (either a mobo jumper, or a BIOS option) to reduce it to 100. Also oddly, I tried a "slower type of CPU": a T-Bird 1400 133FSB CPU, and the mobo set the FSB to 100, with no option to increase the FSB to 133. I think that the BIOS is the latest: April 2003.

I could uninstall the 4.43 VIA chipset driver, and either try an earlier version (for example 4.35), or perhaps a later (last?) version, eg 4.56. One other option would be to uninstall the Breese latency patch, but I've heard good things about it, and so I don't really want to do this.

I've just removed a basic 2MB PCI graphics card, and installed a Ti4200 AGP graphics card, and by doing this, the sound delays can be heard all the time now. It doesn't matter if the CPU speed is set to 800mhz or 1400mhz or higher, I always get sound delay problems.

retro games 100 wrote:

I now need to get some "electric conductive insulator", in order to close L5 [1]. (Do I also need to close L5 [2], in order to set "mobile mode" to "true"?) I tried looking on ebay for electric conductive insulator (and similar phrases), but my searches showed no results.

OMG. 🤣. This post made my day.

"electric conductive insulator" << read that and think about it.

What I have read is that you can get a rear window defroster repair kit, it comes with a conductive paint that should work well. I'm also pretty sure the L5's don't have the pits etched between the contacts so you don't need to fill anything in like the L6. If your chip is a mobile version, you shouldn't need to change L5[2], it's already the way you want it.

Regarding the sound issue, my memory is a little fuzzy, but the only issue I remember having with my ISA sound had to do with DOS and midi, and the main issue there was my lack of effort in getting the config.sys correct.. I had a similar idea about the voltage drop pop into my head while making my coffee, I hadn't even read that part of the post yet. I guess you could use speed fan or some other utility to monitor the 5v while you change multi's and engage the sound card.

Edit: try loading your bios failsafe defaults, lets see if that has any effect.

When this thread was started, I had my secondary case built up with a different system. I wanted to have some fun and games too, but after discovering this I had to hit the brakes:

As you can see, the goo has busted out the bung and leaked across several circuits 😲 🙁

Strangely enough my system was at it's peak of stability when it was dis-assembled, but still I'm wary of putting back in this condition. That cap had always been very bulged, and I guess a thousand or so hours of constant use pushed it over the edge.

prophase_j wrote:

retro games 100 wrote:

I now need to get some "electric conductive insulator", in order to close L5 [1]. (Do I also need to close L5 [2], in order to set "mobile mode" to "true"?) I tried looking on ebay for electric conductive insulator (and similar phrases), but my searches showed no results.

OMG. 🤣. This post made my day.

"electric conductive insulator" << read that and think about it.

RG100 thinks a bit...then gets a headache. 😉

Vaguely in my "defence", it says on that website:

How to set the bridge of [Open] to [Close]
We know the Laser pit should be filled with an insulator before anything else. Then, it is made to [Close] by using electric conductive material.

It's very unfortunate that the cap goo has burst on your mobo. I am fortunate, because the bulging caps on my board have not yet burst.

BTW, I tried setting all the BIOS options to "fail safe" values, but that didn't fix the delay sound problem. I also tried 2 other AGP cards, an ATI 7500 and a Voodoo 3. No joy. I did try uninstalling the Breese latency patch, but didn't help.

I might try a few other VIA 4-in-1 driver versions. Also, I wish I could locate my AWE64 card. If that works OK, I would be happy with that. I could use the EWS sound card in another mobo.