VOGONS


First post, by WJG6260

User metadata
Rank Member
Rank
Member

Hello all,

Tl;dr: Bold text below contains the root of the issue and what I've discovered so far...

Recently, I was working on an upgrade for my Pentium III build and decided to swap the motherboard and upgrade the CPU.

Previously, my machine had the following specifications:

  • ASUS CUV4X
  • Pentium III 1.0EB Coppermine-T
  • 512MB PC133 SDRAM CAS3
  • 3dfx Voodoo 3 3000 AGP
  • Intel 10/100 Network Card
  • Aureal Vortex 2 (TB Montego II)
  • SoundBlaster AWE 64 Gold

It was a pretty nice machine, but I had a few problems with it. Although this pre-dates the issue I'm having, I suppose it's worth mentioning that I switched to the CUV4X in an attempt to accomplish two things: (i) pass the 512MB RAM limitation of the intel 815 chipset; and (ii) make sure that I have a board suitable for hosting my AWE64 Gold as a secondary MIDI card.

First, I found that there definitely was a slowness to the GUI that I hadn't noticed when using an ASUS TUSL2-C with the same Voodoo 3 3000 AGP. I came across this interesting thread and it seems that I am not the only one with this issue. Normally, I would not care, however I found Windows 98SE to be sluggish on both a fast CF card and SSD, and decided that enough was enough. The machine was stable, however, and seemed to provide pretty good performance in UT99, which I find myself playing quite a bit.

In an attempt to raise the bar a bit and put together a more performant machine, I purchased recently an ASUS CUBX-E, thinking that a 440BX board at 133MHz would be the way to go; 89MHz AGP aside, I figured that the board overall would offer more versatility and far higher performance. I decided to mate this with a Pentium III 1.26-S from the Korean eBayer who sells pin-modded CPUs, and used the BIOS a fellow member of this forum produced to ensure good compatibility with the board.

The machine today sits with a repasted video card known to work acceptably with 89MHz AGP, an SSD on one of those bidirectional SATA-IDE converters with the Satalink Spif223a chipset, and the following overall set of specifications:

  • ASUS CUBX-E
  • Pentium III 1.26-S Tualatin (pin-modded)
  • 512MB PC133 ECC SDRAM CAS3
  • Nvidia GeForce 4 Ti 4200 64MB
  • 3COM 3C905B 10/100 Network Card
  • Aureal Vortex 2 (TB Montego II)
  • SoundBlaster AWE64 Gold

I figured that the 440BX chipset would be the answer here, being stable, fast, and proven. This CUBX-E worked just fine with a Penitum III 667MHz, a Pentium III 1.0EB, and the Coppermine-T from above. With the new BIOS, it was just fine with the Tualatin too--except in Windows, wherein I find the machine randomly freezing. The machine will lock up most of the time when sound on the Vortex 2 is being played and something intensive is being done on the SSD--my thoughts immediately went to there being an IRQ issue. Sometimes it'll lock up even if sound isn't playing but if I'm, say, installing a game using Daemon Tools whereby the sound card should be playing something at the menu.

I've had numerous issues with IRQs and presumed that there's something going on there: it appears that the network card was at one point conflicting with the IRQ for the Promise ATA100 controller on the board, and at another point it appears that the Vortex 2 was in conflict with the Promise controller. I swapped slots around, and now the Promise controller has IRQ10 all by its lonesome and the Vortex 2, 3COM card, and USB controller all share IRQ 9.

I've tried all I can think of--different slots, reinstalling Windows, reconfiguring drivers, a different BIOS, different RAM (although the ECC sticks used here are known good that I've had for a long time and used in a few other builds over the years), and lots of patience and hair-pulling and nothing has changed. I know the cards from the original build are all good and the board looks to be in good shape with no visible trace damage or anything; I tested it in various configurations, as mentioned above, and it seemed okay.

At this point I'm about ready to give up on the CUBX-E, but I really want an ISA slot or two in my Pentium III build. I've considered a Pentium 4 1.5GHz build with an ASUS P4T, but I don't want to yield on being able to use the AWE64 Gold just yet. I also have an FIC KC19+ and it's a nice board, with the weird but pretty decent Intel 820 chipset, however, the ISA slots don't have working DMA, as far as I can tell. Firing up DOOM or DOOM 2 doesn't give me any sound effects, neither does Wolf3D.

Part of me wishes I could find something like the ASUS P3C-E or one of its brethren, as I'm totally fine with RDRAM, and would be interested in having working ISA (by ITE8888F) and 133MHz FSB without overclocked AGP on an Intel chipset, but the rational side of me says that it's not worth going on a wild goose chase for one particular board that's pretty darn rare and with a chipset that's potentially not much better than the Apollo Pro 133A (nor is it worth the $$$ 🤣).

I suppose the question then comes down to this: Has anyone faced anything similar and, if so, what did you do to remedy the issue? I'd really appreciate any and all suggestions!

I'm sorry this is such a long post, but I figured I'd provide as much background as possible. If there's any more info y'all need to help figure this out, please let me know.

And a massive thank you in advance to all those who respond and provide some insight!

-Live Long and Prosper-

Feel free to check out my YouTube and Twitter!

Reply 1 of 6, by flupke11

User metadata
Rank Oldbie
Rank
Oldbie

As the board is overclocked, do the same issues occur when the board is set at default speed with default bios? I know that you want it to run at 133, but I would suggest running it plain vanilla first to find the root cause.

Reply 2 of 6, by WJG6260

User metadata
Rank Member
Rank
Member

Thanks for getting back to me flupke11!

I gave your suggestion a go before heading to bed. Things seemed a little more stable at first, but the same dang thing happened again when heavy disk I/O and the Vortex 2 were used at the same time. I can't exactly figure out why, as I made certain they're on different IRQs (the Promise controller is on 10 and the Vortex is on 9). I'm almost wondering if for some reason this board is just being stubborn with the Vortex? PCI isn't overclocked here, as this has the proper 33.25Mhz divider, but the thing that I can't stop thinking is that there's either some sort of hardware issue or resource conflict and being coupled with the 133MHz FSB things just aren't happy.

I also pulled everything and stuck it back in the CUV4X, and there everything seemed perfectly stable. I did notice that there was less IRQ sharing going on as well.

I'll have to keep at this, as I'm not giving up just yet! Thank you so much again for the help though, I hadn't thought of that and it almost seemed like the answer! Dang this finnicky old hardware...

-Live Long and Prosper-

Feel free to check out my YouTube and Twitter!

Reply 3 of 6, by WJG6260

User metadata
Rank Member
Rank
Member

Alright, slight update here:

I just used Riva Tuner to disable Side Band Addressing on the off-chance that there was an issue with this feature. At AGP 2X and with SBA off, everything's been stable so far. Strange, but I have read that this issue did plague some overclocked 133MHz 440B machines back in the day and do recall that it was common to have to disable SBA. I thought this only really affected earlier GeForces though? Strange.

Either way, so far, so good!

-Live Long and Prosper-

Feel free to check out my YouTube and Twitter!

Reply 4 of 6, by AlexZ

User metadata
Rank Member
Rank
Member

I would recommend trying Pentium III 1.0EB instead of the Tualatin for a while, at 133 and 100 FSB before giving judgement it's caused by SBA. I would also recommend removing unnecessary expansion cards and drivers. A bug in driver can crash the whole system.

I would always take ASUS CUBX-E over CUV4X as 440BX is faster. I have a few 440BX boards but no VIA.

Pentium III 900E, ECS P6BXT-A+, 384MB RAM, NVIDIA GeForce FX 5600 128MB, Voodoo 2 12MB, 80GB HDD, Yamaha SM718 ISA, 19" AOC 9GlrA
Athlon 64 3400+, MSI K8T Neo V, 1GB RAM, NVIDIA GeForce 7600GT 512MB, 250GB HDD, Sound Blaster Audigy 2 ZS

Reply 5 of 6, by WJG6260

User metadata
Rank Member
Rank
Member

Thanks for the insight AlexZ! I'll give that a go as well. Do you think it matters if the 1.0EB is a Coppermine-T or a standard model? From what I understand there should not be any real differences as far as the 440BX setup is concerned, right?

I do still think this could be IRQ related, but my GF4 Ti 4200 is also really the only card I can test at the moment at 89MHz. I know for a fact that it is rock stable at 100MHz, having used it prior to this in a KT133 board for a while.

I'll give your suggestions a go and see what differences in stability there are. I didn't think much about drivers, but you've got a great point there--I know that for the GF4 the recommended driver is often 45.23. Is there any other driver set that might be worth looking at?

Thanks again for all the help!

-Live Long and Prosper-

Feel free to check out my YouTube and Twitter!

Reply 6 of 6, by AlexZ

User metadata
Rank Member
Rank
Member

I would recommend standard PIII 1Ghz first and if it's stable at 100 and 133 FSB then work your way up.

Pentium III 900E, ECS P6BXT-A+, 384MB RAM, NVIDIA GeForce FX 5600 128MB, Voodoo 2 12MB, 80GB HDD, Yamaha SM718 ISA, 19" AOC 9GlrA
Athlon 64 3400+, MSI K8T Neo V, 1GB RAM, NVIDIA GeForce 7600GT 512MB, 250GB HDD, Sound Blaster Audigy 2 ZS