I have attached the Jan Steunebrink BIOS patch for Cyrix 4x, AMD K6-2, AMD K6-2+, AMD K6-III, AMD K6-III+, and 128 GB HDD support and all BIOSes I have for this board. See BIOS.txt in the zip file. When using Jan's BIOS, I recall that one or both of system/video BIOS cacheable had to be disabled. I forget which now, so test this.

I've attached the jumper settings from the Biostar website, as well as that from Stason. I've also attached the pdf for the clock generator chip on this motherboard.

If you plan to run it with a 75 MHz or 83 MHz FSB, I have found that 256 MB of SDRAM may not be that stable. So for 75 or 83 MHz, if you insist on using SDRAM, you may need to use only 64 MB. On the other hand, I have had no problem running 256 MB of EDO 72-pin SIMMs at 83 MHz. On this board, I have not noticed any performance penalty when using EDO over SDRAM. At 83 MHz (and maybe 75 MHz?), if you are using an AMD K6-III, K6-III+, or K6-2+, you will need to disable the motherboard's L2 cache for it to run stable. But, you cannot simply disable it in the BIOS. In fact, the contrary is true - you need to leave L2 enabled in the BIOS, but must remove the DIP-28 TAG RAM chip to disable L2 cache. Some BIOS or PCB design issue? I'm not sure, but removing the TAG RAM solved the problem.

EDIT: for runnig an AMD K6-III+ at 500 MHz, you should add a small fan to the motherboard's VRM heatsink. Also, use the measured 2.02 Vcc2 jumper setting I noted in the undocumented jumper table. My system is quite stable, stable enough to run XP Pro SP3 and run all the Windows updates for days on end at 100% CPU usage. Don't forget the VRM fan though.

Much appreciated..
This build will probably be either win95 or win3.1 ...for some win gaming plus running dos games in pure dos mode.
I've sorted out a creative SB16 pnp for it (CT2980)
what I dont have is a decent ISA graphics card ..though hopefully I'll be able to install a PCI card to do the graphics.
Not sure if I'd then need DOS drivers ???
3x ISA slots.
here is what the ebay listing said ..
and what is included ...

"Vintage Pentium MMX socket 7 with motherboard. MB-8500TTD. Excellent Condition.. Condition is Used.

This was working with windows 95 before being removed from the PC case. Would be a nice challenge to get this working again with windows 95, and the required peripherals."
rgds
VS

Hmmm I wonder..??
Perhaps this is finally an opportunity to use my PCI voodoo 5 5500 I repaired last year..

feipoa wrote:

I have attached the Jan Steunebrink BIOS patch for Cyrix 4x, AMD K6-2, AMD K6-2+, AMD K6-III, AMD K6-III+, and 128 GB HDD suppor […]
Show full quote

I have attached the Jan Steunebrink BIOS patch for Cyrix 4x, AMD K6-2, AMD K6-2+, AMD K6-III, AMD K6-III+, and 128 GB HDD support and all BIOSes I have for this board. See BIOS.txt in the zip file. When using Jan's BIOS, I recall that one or both of system/video BIOS cacheable had to be disabled. I forget which now, so test this.

I've attached the jumper settings from the Biostar website, as well as that from Stason. I've also attached the pdf for the clock generator chip on this motherboard.

If you plan to run it with a 75 MHz or 83 MHz FSB, I have found that 256 MB of SDRAM may not be that stable. So for 75 or 83 MHz, if you insist on using SDRAM, you may need to use only 64 MB. On the other hand, I have had no problem running 256 MB of EDO 72-pin SIMMs at 83 MHz. On this board, I have not noticed any performance penalty when using EDO over SDRAM. At 83 MHz (and maybe 75 MHz?), if you are using an AMD K6-III, K6-III+, or K6-2+, you will need to disable the motherboard's L2 cache for it to run stable. But, you cannot simply disable it in the BIOS. In fact, the contrary is true - you need to leave L2 enabled in the BIOS, but must remove the DIP-28 TAG RAM chip to disable L2 cache. Some BIOS or PCB design issue? I'm not sure, but removing the TAG RAM solved the problem.

EDIT: for runnig an AMD K6-III+ at 500 MHz, you should add a small fan to the motherboard's VRM heatsink. Also, use the measured 2.02 Vcc2 jumper setting I noted in the undocumented jumper table. My system is quite stable, stable enough to run XP Pro SP3 and run all the Windows updates for days on end at 100% CPU usage. Don't forget the VRM fan though.

Do you think there is a chance that a faster TAG RAM chip would work at 83 MHz? Or would it even be worth the bother?

If you are using a K6-III or K6-2/3+ chip, you'll need to remove the TAG RAM for reliable operation.

Here's the photos I said I'd deliver.

First is, remove the TAG RAM. Notice the empty DIP. And use 72-pin EDO, not SDRAM.

Here's the fan needed for the VRM. I was able to screw it into the fin gap on the onboard VRM heatsink with just one screw. VRM stays cool, even with just a few CFM of air flow.

Thanks for replying. The system is working fine @ 450 MHz with 256 megs of SDRAM, but I still have the TAG installed, and it won't even post when set to 83 MHz FSB. So my question was if you thought it might be possible that the issue is that the TAG chip itself is too slow for anything over 75 MHz?

If you think its stable, try installing XP and running all the XP updates. LOL! For 256 MB of SDRAM to really be stable, you'd have to run it at 400 MHz. Or use 72-pin RAM. This is my experience anyway. It appears stable at first, but takes time to really narrow it down.

Changing the TAG won't help. I use a 10 ns TAG which worked fine for chips at 83 MHz which did not have their own L2 cache, like the Cyrix MII. But using the 10 ns TAG with 83 MHz and the K6-III+ did not work. There is almost no speed benefit in using the onboard cache when using an AMD K6-III+ on this motherboard.

feipoa wrote:

If you think its stable, try installing XP and running all the XP updates. 🤣! For 256 MB of SDRAM to really be stable, you'd have to run it at 400 MHz. Or use 72-pin RAM. This is my experience anyway. It appears stable at first, but takes time to really narrow it down.

Changing the TAG won't help. I use a 10 ns TAG which worked fine for chips at 83 MHz which did not have their own L2 cache, like the Cyrix MII. But using the 10 ns TAG with 83 MHz and the K6-III+ did not work. There is almost no speed benefit in using the onboard cache when using an AMD K6-III+ on this motherboard.

Challenge accepted! XP installed fine (albeit slowly) and the updates went pretty well, though there was one wrong turn that was my fault. There was a driver listed for the Radeon 9250 I have installed, and I chose to install it with the updates. As soon as that was installed, I began getting blue screens that were related to that driver, and I had to do a system restore to get rid of it. I then did the updates again, and all is well. But to state the obvious, WinXp is too much for a system this slow, and with this little memory. 3D Mark 2000 gave me a score of 750 with the 233 MMX, and around 1,900 with the K6-2+ @ 450 with Windows 98 SE - that dropped to about 700 points with Windows XP! It really is a shame that the TX chipset is limited to 256 megabytes of memory. This motherboard makes me miss my trusty Asus TXP4, but they are going for over twice what I paid for this on Ebay.

How did you finish installing all Windows XP updates? There are hundreds of them and it took my system more than 2 days.

feipoa wrote:

How did you finish installing all Windows XP updates? There are hundreds of them and it took my system more than 2 days.

I have a Team group 120 megabyte SSD connected to a Promise ATA controller card with a SATA to IDE adapter, so that probably sped things up a bit. It still took over an hour to download them all, and at least that long to install. Starting with SP3, and immediately updating to IE8 and the IE8 update (so I could connect to Windows Update), there were 132 updates to download and install.

Also, some years ago when I had my TXP4, I swapped the 256 megabytes of SDRAM for 256 megs of EDO (I had a set of four 128 megabyte EDO SIMMs to max out the memory of a P55T2P4) and while most things seemed about the same, the hard drive I had attached with a Promise card had significantly lower read/write speeds with the EDO than it did with the SDRAM.

Anyway, I've had fun with it, and I would never have thought this board could work at all with a K6-2+ (or known of the undocumented jumper settings, or found the Jan Steunebrink BIOS) if not for your posts on this message board, so thanks. And that VRM does get hot (even with the 233 MMX), I'll have to do something about that.

Are you going to try 83 MHz? Do you also have a K6-III+?

Does Windows update know to remove updates that require SSE? Or do you know which updates not to install which require SSE? A few years back when I went to update, shortly after updates required SSE, the system would no longer boot up. I'm curious how you got around this.

...the hard drive I had attached with a Promise card had significantly lower read/write speeds with the EDO than it did with the SDRAM.

This is interesting. Did you note any benchmark values.

Might also be a good idea to let MemTest 4.00 run overnight to see if any errors pop up.

I agree, shame about the 256 MB chipset limit. XP SP3 fills that up on boot and struggles with paging while running the Windows update client.

Your CPU-Z image doesn't show that there is any onboard L2 cache. Did you remove the TAG RAM? For some reason, I thought you had kept the L2 cache enabled.

feipoa wrote:

Are you going to try 83 MHz? Do you also have a K6-III+? […]
Show full quote

Are you going to try 83 MHz? Do you also have a K6-III+?

Does Windows update know to remove updates that require SSE? Or do you know which updates not to install which require SSE? A few years back when I went to update, shortly after updates required SSE, the system would no longer boot up. I'm curious how you got around this.

...the hard drive I had attached with a Promise card had significantly lower read/write speeds with the EDO than it did with the SDRAM.

This is interesting. Did you note any benchmark values.

Might also be a good idea to let MemTest 4.00 run overnight to see if any errors pop up.

I agree, shame about the 256 MB chipset limit. XP SP3 fills that up on boot and struggles with paging while running the Windows update client.

Win update must do a system check and not include any updates requiring SSE (just as certain web browsers would refuse to install once they began requiring it) because I just let the client do its thing, and it had no issues. But as I said, it isn't perfect, because it did offer a video card driver that caused BSODs.

When I tested drive speeds years ago, it was with a defrag program called Vopt XP which also had a drive speed tester. Little known, and the drive speeds it would give were only really relevant if you used the same program so you could get an apples to apples comparison. But I did, and I had noticed that the P55T2P4 (which has the HX chipset and therefore is EDO only) had significantly lower drive speeds than the TXP4 I had with all else being equal. So I decided to swap the TXP4's memory, and that showed that it was the memory at fault for the slowdown. The default IDE ports on the P55T2P4 tested at about 10 MBPS with a fast IDE drive (7,200 16 megs of cache) but IIRC only got up to 15 to 20 MBPS when attached to a Promise ATA 100 card - the same drive on the same card in a TXP4 would deliver 30 to 40 MBPS. I also used all of the same on a K6BV3+, and though the IDE port on that board was supposed to be ATA 66, it delivered speeds very comparable to the default ATA 33 ports on the TXP4, and both were about the same when using the Promise card. One thing I definitely remember was that Win98 was a very stable OS on the Intel chipset Socket 7 boards - I was mystified by all of the talk of it being unstable until I tried using Via chipset mobos. But Via chipset mobos were just fine with WinXP in my experience.

I just found that Vopt is now freeware, and the slightly newer version shows this system delivers about 52 MB/second. https://www.majorgeeks.com/files/details/vopt.html

After having no luck setting it to 83 MHz FSB with Windows 98 (and not willing to remove memory or switch to EDO) I'd given up on the idea. But I just gave it another try, and it booted to WinXP with no issue, and it just completed a run of 3D Mark 2000 with a score of 1,005. I wonder why Windows 98 has a problem with this mobo at 83 MHz? I've installed Win98 on both Asus TXP4 and P55T2P4 motherboards with K6-2+ CPUs many times over the years with no issues at all, and would have hoped that the same would be true here? I was ready to blame the ATI video card - I've always heard they're less tolerant of out of spec PCI bus speeds, and 4.71 Mhz is pretty far out of spec. But it's working fine now, so that's not it.

No, I don't have a K6-III+ (I wish I did), they just get called that by programs that recognize a K6 with level 2 cache as a K6-3.

feipoa wrote:

Your CPU-Z image doesn't show that there is any onboard L2 cache. Did you remove the TAG RAM? For some reason, I thought you had kept the L2 cache enabled.

Yeah I had to remove it.

mmm. Perhaps the issue has been fixed with Windows update. I know it busted my K6-III installation. It also messed up some more recent XP POS2009 installations which required a restore after some updates expected SSE2. Recent WinXP/POS2009 update causes errors with PIII systems

When I used NT4 back in the day, I bought VopNT v3.00 to defrag. I wasn't aware there was a speed utility. Would you be willing to benchmark the HDD on your Biostar MB-8500TTD with SDRAM vs. EDO?

I sort of recall 83 MHz not wanting to boot Win98se w/SDRAM but would boot W2K. Switching to EDO RAM solved it for me. At 75 MHz, I thought 256 MB of SDRAM was stable and I had ran it like that for a long time, but little hints here and there caused me to switch to EDO. I think 83 MHz with EDO will be faster than 75 MHz with SDRAM. If you want to stick with SDRAM at 83 MHz, try out just a single stick of 64 MB to see if it works. When pushing the limits, I find that less RAM solves stability issues. This is especially true in this case and when running the Cyrix MediaGX at 300 MHz.

I'd think that a newer card like the Radeon 9250 would be PCI 2.1 compliant, which should allow for up to 66 MHz on the bus. I've run a GF2MX, GF4MX, and FX5500 in mine with a 42 MHz bus without issue.

I left Memtest86 running overnight with the bus set to 83 MHz, no errors.

I have not done the POS hack, maybe that's the difference?

If I had a pair of 128 meg EDO SIMMs on hand, I'd be happy to do the comparison, but I only have 48 megs total on hand.

Repo Man11 wrote:

I left Memtest86 running overnight with the bus set to 83 MHz, no errors.

With a single stick of 128 MB or two? Try just one stick and booting Win98.

Repo Man11 wrote:

I have not done the POS hack, maybe that's the difference?

Most likely. I think all my XP systems are POS2009. If you are feeling adventurous, adapt them to POS2009 and try running updates. Be sure to image the drive first! Don't rely on system restore - it has caused profile errors on my more modern systems such that I'm always using a temporary profile now.

Repo Man11 wrote:

If I had a pair of 128 meg EDO SIMMs on hand, I'd be happy to do the comparison, but I only have 48 megs total on hand.

I'll add it to my list then. But to be honest, my list is so long now that I'll probably never get to it.

Leaving it as it was, I popped another hard drive in it and reinstalled Win98. And it runs fine. I guess removing the TAG RAM after installing 98 caused some sort of issue? While web surfing on my primary computer, I've been running 3D Mark 2000 over and over, with no issue, score is in the 2,000+ range now. it's definitely happier with Win98 than it was with XP.

Right now it has one 512 megabyte stick of PC133, which the chipset only sees half of.