VOGONS


First post, by WJG6260

User metadata
Rank Member
Rank
Member

Hello friends, it’s sure been some time! I hope all of you have been well!

I acquired, by chance and luck, two Socket 5 PCI-VLB-ISA boards. One of them, the SuperMicro P54VL-PCI, has appeared on this forum a few times.

The first is a Northgate Elegance Pentium 90, equipped with the OPTi 546/547 “Python” chipset. It’s a solid board and I find two things about it quite nice: (i) the asynchronous VLB bus clock is modifiable by jumpers; and (ii) the CPU bus clock is fully adjustable between 50MHz and 66MHz. There’s probably other undocumented settings, but I haven’t messed with the board much, as I need to replace the lithium Tadiran CMOS battery. The build quality of this thing is solid, but it’s not the best. I find the chipset to be quite slow, which I gather is to be expected from these OPTi Pentium/VLB implementations. What’s nice, though, is that I can run an IDT WinChip2 at 50X4, and get pretty solid performance at 200MHz.

That leaves me with the other board, which I find far more interesting: the SuperMicro P54VL-PCI. It’s a much nicer board, and uses the OPTi 546/597 chipset. I haven’t seen any other boards laid out with part of the 546/547 chipset and part of the 596/597 “Cobra” chipset. Before my actual question, does anyone have any clue why this could be implemented as such? I’ve read something that indicates that the OPTi Python had some corrections/upgrades over the Cobra, but I’m not sure about that. The SuperMicro P54VL-PCI seems to provide MUCH higher memory bandwidth and throughput; no single benchmark that I’ve ran has turned the tide towards the other board and its OPTi Python chipset.

This leads me to my question and predicament. The SuperMicro P54VL-PCI is awesome, but it only has a single jumper to select between a 60MHz and 66MHz FSB. That means that the WinChip2 would be running at 240MHz with a 60MHz bus, and my chip does not seem happy there. It appears there are Pentium OverDrives in both P54C and P55C flavors for this situation, but are they the only option? If I were to go that route, would the P55C/Pentium OverDrive MMX work on this board? The VRM seems beefier than that on the Northgate, but I haven’t really read much into its specifications.

If anyone has any experience with something like this, I would love to hear their thoughts. I’ve attached pictures of the boards, and can provide more information if needed.

Thank you again for the help and insight!

Attachments

Last edited by WJG6260 on 2021-09-09, 21:44. Edited 1 time in total.

-Live Long and Prosper-

Feel free to check out my YouTube and Twitter!

Reply 1 of 20, by jheronimus

User metadata
Rank Oldbie
Rank
Oldbie

I also own a P54VL-PCI and an Opti 546/547 board, Biostar MB-1500PCT and I've spend almost a year restoring a tinkering with them. I absolutely love this obscure hardware, and I'm very slowly working towards a finished build and a proper comparison with Intel Zappa (basically the board that killed quite a few of 3rd party chipset makers including Opti). I think I'll make a big post once I find a good case for it — you can tell it's kind of a passion project for me 😀

The first is a Northgate Elegance Pentium 90, equipped with the OPTi 546/547 “Python” chipset

Could you please upload the BIOS for this board? I know only of two BIOS images for this chipset — the one for Edom/WinTech MP039 and mine, and my board certainly doesn't like the WinTech BIOS.

Also, what version of BIOS are you running on Supermicro? Do you by any chance have the manual for it? According to some Usenet posts I've seen it does exist, but all I could find was the Total Hardware 99 page that has like 75% of jumpers listed as unknown.

What’s nice, though, is that I can run an IDT WinChip2 at 50X4, and get pretty solid performance at 200MHz.

It's interesting, because I've tried the chip on both boards and it didn't POST. I'm starting to think the chip is actually dead. Don't think I've ever gotten a dead CPU of that generation, so I never thought to check with a different board. What's the make of your chip? Mine is C6-PSME200GA.

It’s a much nicer board, and uses the OPTi 546/597 chipset.

Cool, I didn't know that! My 546 chip is covered with a sticker that I didn't want to remove.

I’ve read something that indicates that the OPTi Python had some corrections/upgrades over the Cobra, but I’m not sure about that.

Some people on Usenet did claim that Python is a "fixed" revision of Cobra, there's very little difference between the two, and I think that most of it is up to the motherboard manufacturer.

My Biostar board can endure the tightest cache and RAM timings (e.g. 3-2-2-2/3-2-2-2 vs 3-2-2-2/4-2-2-2 on Supermicro) and supports more cache (2MB vs 512KB on Supermicro). On the other hand, Biostar can't handle EDO RAM. Technically, Biostar has a 2/3 divider for VLB that results in 44 MHz speed, but other than some incompatibilities with disk controllers I haven't found any difference.

Overall, the Biostar performs just a bit faster than Supermicro (around 1 FPS difference in most benchmarks), but do keep in mind that I have 1MB of cache on it. The Supermicro board is somehow a bit more stable in my experience, and also has less issues with PCI (I still can't get the NIC to work on the Biostar).

I've tried Pentium MMX Overdrive, WinChip and Evergreen Spectra (AMD K6-2 400). I think I've been able to run Supermicro with both Overdrive and AMD K6-2, but I did have to use BIOS files from other boards. I'll have to look for specific results, but I think I've had luck with Shuttle HOT-423 and some other Award BIOSes. Biostar doesn't boot with anything but Pentiums. It also treats Pentium MMX as a 486 and the performance does suffer.

UPD: Found my old thread on upgrade chips for the Supermicro. Yeah, I did use a BIOS from the TMC board to run a Pentium MMX Overdrive.

I think the safest option would be a regular Pentium — either in a form of an Overdrive or using a CPU interposer. I haven't tested this yet, but at least Supermicro does boot with Pentium 200 — even if it treats it as a Pentium 100. I have a hunch that the Opti chipsets need to tell Pentium from 486, so many BIOS versions somehow check for it.

Last edited by jheronimus on 2021-09-07, 22:19. Edited 2 times in total.

MR BIOS catalog
Unicore catalog

Reply 2 of 20, by jheronimus

User metadata
Rank Oldbie
Rank
Oldbie

The fact that Supermicro uses a "mixed" chipset would explain why PC Magazine says this in their review

Screenshot 2021-09-07 at 23.50.33.png
Filename
Screenshot 2021-09-07 at 23.50.33.png
File size
163.73 KiB
Views
1405 views
File license
Public domain

Note that almost all Pentium VLB boards in their review are Supermicro P54VL-PCI.

MR BIOS catalog
Unicore catalog

Reply 3 of 20, by WJG6260

User metadata
Rank Member
Rank
Member

Thanks for the reply and for your insight! I really do appreciate your thoughts on this!

jheronimus wrote on 2021-09-07, 20:46:

I also own a P54VL-PCI and an Opti 546/547 board, Biostar MB-1500PCT and I've spend almost a year restoring a tinkering with them. I absolutely love this obscure hardware, and I'm very slowly working towards a finished build and a proper comparison with Intel Zappa (basically the board that killed quite a few of 3rd party chipset makers including Opti). I think I'll make a big post once I find a good case for it — you can tell it's kind of a passion project for me 😀

Agreed on this one! You can't beat these weird Pentium VL-Bus hybrid boards. There's something about them that-while objectively confusing and perhaps not the best combination for performance and stability-works wonders in a way that makes using them a treasure! I can't wait to see your big post about this comparison; the Zappa is pretty formidable. 430FX is an underrated chipset, and offers a massive upgrade over earlier chipsets.

Could you please upload the BIOS for this board? I know only of two BIOS images for this chipset — the one for Edom/WinTech MP039 and mine, and my board certainly doesn't like the WinTech BIOS.

Done! I spent much of yesterday dumping the BIOS from this board and testing it. The Phoenix BIOS it comes with is acceptable, but performance seems somewhat underwhelming. If it wasn't for that fact, I would probably grab it before I did the SuperMicro board. By the way, I tried tighter cache timings on the Northgate board and, lo-and-behold, it turns out that you can run MUCH tighter RAM and cache timings on this thing. I wonder if the SuperMicro board is limited because of the chipset, or because of the BIOS?. Did you test the board with tighter timings on that TMC BIOS?

Also, what version of BIOS are you running on Supermicro? Do you by any chance have the manual for it? According to some Usenet posts I've seen it does exist, but all I could find was the Total Hardware 99 page that has like 75% of jumpers listed as unknown.

I was using version 3.0. I uploaded that, but, while dumping the BIOS ROM, I flashed it to 3.2, available here: http://www.win3x.org/uh19/motherboards/4759.

As for the manual, I do have the manual! In fact, the manual I have is also for a P5VL-PCI Socket 4 board. I've scanned and uploaded it to my Google Drive, since the file is quite big. Here's a link: https://drive.google.com/file/d/1ZBYWiR3X_-_f … iew?usp=sharing Hopefully it serves you well.

It's interesting, because I've tried the chip on both boards and it didn't POST. I'm starting to think the chip is actually dead. Don't think I've ever gotten a dead CPU of that generation, so I never thought to check with a different board. What's the make of your chip? Mine is C6-PSME200GA.

I was able to run both a WinChip and WinChip2. The WinChip is a C6-PSME200GA, and the WinChip2 is a W2-2DEE200GSA. Interestingly, the WinChip2 at 200MHz was pretty fast. It ran much better than I expected with a 50MHz bus. Unfortunately, either due to VRM or chip limitations, the SuperMicro will boot at 240MHz on both of these, but will not POST. That's precisely why I'm considering some sort of OverDrive. But, I think that I'm going to repair the CMOS battery on the Northgate before deciding which board to put into a case and mess with.

Cool, I didn't know that! My 546 chip is covered with a sticker that I didn't want to remove.

Now this is the weird thing: I was reading the manual while scanning it, and SuperMicro indicates that the board SHOULD have a 596/597 chipset. As to whether I have a later, "fixed" revision of the board, or if perhaps the 546 and 596 are interchangeable, I am not certain, but the fact remains that there is reason to suppose both could be true. It seems like there's so little information out there on these boards. It's a shame, really, as they're super cool!

My Biostar board can endure the tightest cache and RAM timings (e.g. 3-2-2-2/3-2-2-2 vs 3-2-2-2/4-2-2-2 on Supermicro) and supports more cache (2MB vs 512KB on Supermicro). On the other hand, Biostar can't handle EDO RAM. Technically, Biostar has a 2/3 divider for VLB that results in 44 MHz speed, but other than some incompatibilities with disk controllers I haven't found any difference.

Interesting. I have not tested my SuperMicro with EDO, but I do notice that the Northgate board will not boot with EDO RAM. The Northgate has a stock 33MHz asynchronous VLB clock, but that is adjustable up to 50MHz. It was screwy at 50MHz though; and that was with a few cards rated for 50MHz (ARK1000VL, S3 Trio32, Cirrus Logic GD5426).

Overall, the Biostar performs just a bit faster than Supermicro (around 1 FPS difference in most benchmarks), but do keep in mind that I have 1MB of cache on it. The Supermicro board is somehow a bit more stable in my experience, and also has less issues with PCI (I still can't get the NIC to work on the Biostar).

This I find fascinating. My Northgate board performs objectively worse in memory-related metrics, and I'm not sure why. It could be that Phoenix BIOS, but I'm doubtful. Perhaps there's something I'm missing?

That being said, the PCI problems did affect my Northgate board as well. I cannot get the board to boot with a PCI video card and display an output. It will boot, but there is no video displayed on the screen. The board does work fine with VLB and ISA cards, and that's fine. I figure that it's something related to the IRQ configurations and the Edge/Level interrupt trigger settings in the BIOS. I have yet to confirm that theory, but it appears the SuperMicro can handle resource allocation on PCI cards much better; again, that's possibly attributable to that Phoenix BIOS, but I am not sure.

I've tried Pentium MMX Overdrive, WinChip and Evergreen Spectra (AMD K6-2 400). I think I've been able to run Supermicro with both Overdrive and AMD K6-2, but I did have to use BIOS files from other boards. I'll have to look for specific results, but I think I've had luck with Shuttle HOT-423 and some other Award BIOSes. Biostar doesn't boot with anything but Pentiums. It also treats Pentium MMX as a 486 and the performance does suffer.

UPD: Found my old thread on upgrade chips for the Supermicro. Yeah, I did use a BIOS from the TMC board to run a Pentium MMX Overdrive.

Huh, wild! I wonder why it treats the MMX as a 486? Perhaps it's some weird chipset quirk. I've been considering a 166MMX or the 180MMX (which I believe is the fastest Socket 5-rated CPU?), if I can find one of either at a reasonable cost. I'll have to mess with other BIOSes. Did you note any real instability? I saw that Quake would not run. I suppose that's forgivable, as long as the board can boot into Windows 95/NT without an issue?

I think the safest option would be a regular Pentium — either in a form of an Overdrive or using a CPU interposer. I haven't tested this yet, but at least Supermicro does boot with Pentium 200 — even if it treats it as a Pentium 100. I have a hunch that the Opti chipsets need to tell Pentium from 486, so many BIOS versions somehow check for it.

I agree with you on this. Interesting that the SuperMicro won't boot with a Pentium 200. I wonder why that is? I think you're right about that. There's something about this lineage of Pentium VLB chipsets that relates to 486 compatibility, as the OPTi 571/572 does come on some weird Socket 3/Socket 4 hybrid boards. I wonder why the WinChips work on my other board. Perhaps that Phoenix BIOS is more forgiving?

I have seen that a MR-BIOS exists somewhere for these boards. Hopefully it's found soon, as I could imagine that that would solve many of our troubles. The locked 1.5X multiplier is unfortunate on these boards, but it's not too tough to deal with.

Attachments

-Live Long and Prosper-

Feel free to check out my YouTube and Twitter!

Reply 4 of 20, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

In theory the best upgrade option for these boards is an Evergreen spectra-400, with a K6-III+ instead of K6-2.

Too bad my VLB/EISA Pentium setup (OPTi chipset) doesn't accept anything but 80501, 80502 and overdrive CPUs.
VLB Pentiums are indeed really cool.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 5 of 20, by WJG6260

User metadata
Rank Member
Rank
Member
Anonymous Coward wrote on 2021-09-09, 15:22:

In theory the best upgrade option for these boards is an Evergreen spectra-400, with a K6-III+ instead of K6-2.

It’s funny you mention that. I was looking around and seem to find that these are the fastest Socket 5-compatible overdrives. Unfortunately, it seems like there are few out there. Even the Pentium OverDrive with MMX seems tougher to find than it should be; I find this confusing, as presumably many were made? It seems like there’s plenty of Socket 3 PODs available, but I’ve had one for years, and that’s not too helpful in this case haha.

Too bad my VLB/EISA Pentium setup (OPTi chipset) doesn't accept anything but 80501, 80502 and overdrive CPUs.
VLB Pentiums are indeed really cool.

I love your VLB/EISA Pentium setup; you’ve got a really a neat backplane/board, but it’s unfortunate that the Acer BIOS is not great. I can sympathize, as the Northgate board I’ve got has a really bad BIOS compared to how nice of a board it is.

Out of curiosity, I haven’t seen the chipset on your Pentium VLB setup used elsewhere—was it designed for active backplane operation?

Either way, VLB Pentiums are underrated setups, surely. They’re not the most performant, no, but they are really fun. It’s a shame that there aren’t many more boards out there with this broad-ranging compatibility. I understand that VLB was a dying bus standard by the time they came around, but there’s plenty of merit to having been able to keep your VLB Windows accelerator or SCSI card, and drop in a Pentium CPU. Too bad the implementation was not the greatest. I’d love to run a VLB Pentium as my tweener Socket 5/7 setup. It’s far more interesting than yet another ASUS T2P4 or 430TX or SS7 setup.

-Live Long and Prosper-

Feel free to check out my YouTube and Twitter!

Reply 6 of 20, by jheronimus

User metadata
Rank Oldbie
Rank
Oldbie
WJG6260 wrote on 2021-09-09, 14:34:

Thanks for the reply and for your insight! I really do appreciate your thoughts on this!

Thank you so much for the manual — this is simply an insane find! At a quick glance it doesn't offer much insight into those jumpers (I was hoping there's a VLB speed divider somewhere), but I'm sure it will give some useful info.

By the way, I tried tighter cache timings on the Northgate board and, lo-and-behold, it turns out that you can run MUCH tighter RAM and cache timings on this thing. I wonder if the SuperMicro board is limited because of the chipset, or because of the BIOS?. Did you test the board with tighter timings on that TMC BIOS?

I believe both boards are very picky about the RAM you use. I ran memtest on my Biostar with a set of FPM sticks that I used on a lot of my 486 boards with no issues and it simply gave me a ton of errors. So maybe I should run the same test on Supermicro and I could use tighter timings with a different set of RAM.

I can't remember the specifics of using alternate BIOSes, but I remember that in general they were slower than the original Supermicro and somewhat less stable. I was having high hopes for the Biostar boards, because I've thought that Award boards didn't block the non-Pentiums, but it's not the case with Biostar, unfortunately.

It's interesting, because I've tried the chip on both boards and it didn't POST. I'm starting to think the chip is actually dead. Don't think I've ever gotten a dead CPU of that generation, so I never thought to check with a different board. What's the make of your chip? Mine is C6-PSME200GA.

Huh, guess I might have a dead WinChip after all, I should test it on a more common board.

Now this is the weird thing: I was reading the manual while scanning it, and SuperMicro indicates that the board SHOULD have a 596/597 chipset.

It even says the same on their website.

As to whether I have a later, "fixed" revision of the board, or if perhaps the 546 and 596 are interchangeable, I am not certain, but the fact remains that there is reason to suppose both could be true.

Interestingly both our boards are rev.3, but yours has these red elements (I don't know what they're called) while mine doesn't. Don't see any other difference in layout.

It seems like there's so little information out there on these boards. It's a shame, really, as they're super cool!

If I am not mistaken, few, if any of these boards were sold in retail, most of them were only available via small system builders. Again, according to Usenet, Supermicro didn't really provide user support — it's a miracle they even released a BIOS update or even acknowledged the board on their website. Then again, it seems to be one of their first products ever (the company started in 1993, a year before this board's time).

The Northgate has a stock 33MHz asynchronous VLB clock, but that is adjustable up to 50MHz. It was screwy at 50MHz though; and that was with a few cards rated for 50MHz (ARK1000VL, S3 Trio32, Cirrus Logic GD5426).

Oh, that's a neat feature anyways.

This I find fascinating. My Northgate board performs objectively worse in memory-related metrics, and I'm not sure why. It could be that Phoenix BIOS, but I'm doubtful. Perhaps there's something I'm missing?

I'll have to get back to you with more numbers for that.

That being said, the PCI problems did affect my Northgate board as well. I cannot get the board to boot with a PCI video card and display an output. It will boot, but there is no video displayed on the screen. The board does work fine with VLB and ISA cards, and that's fine. I figure that it's something related to the IRQ configurations and the Edge/Level interrupt trigger settings in the BIOS. I have yet to confirm that theory, but it appears the SuperMicro can handle resource allocation on PCI cards much better; again, that's possibly attributable to that Phoenix BIOS, but I am not sure.

It's weird that you can't use PCI video (I didn't have that issue on either board). I think WaybackTech had the same exact problem with his TMC Socket 5 board, but he mentions he used a non-original BIOS. His board used a 571/572 chipset and acted very strangely in many regards and I think he had memory bandwidth that was even below 486 levels. So yes, it could be related to BIOS.

Huh, wild! I wonder why it treats the MMX as a 486? Perhaps it's some weird chipset quirk. I've been considering a 166MMX or the 180MMX (which I believe is the fastest Socket 5-rated CPU?), if I can find one of either at a reasonable cost. I'll have to mess with other BIOSes. Did you note any real instability? I saw that Quake would not run. I suppose that's forgivable, as long as the board can boot into Windows 95/NT without an issue?

I don't think any regular MMX chip is rated for 3.3V. I haven't tried Win95 on Supermicro at all — only on Biostar, so can't comment there. But generally compared to DOS/Win3.11, Win9x is a lot more sensitive once you start picking tighter timings.

I have seen that a MR-BIOS exists somewhere for these boards. Hopefully it's found soon, as I could imagine that that would solve many of our troubles. The locked 1.5X multiplier is unfortunate on these boards, but it's not too tough to deal with.

That would be really cool. I've been looking for an image for a very long time in many places 😀

MR BIOS catalog
Unicore catalog

Reply 7 of 20, by jheronimus

User metadata
Rank Oldbie
Rank
Oldbie
Anonymous Coward wrote on 2021-09-09, 15:22:

In theory the best upgrade option for these boards is an Evergreen spectra-400, with a K6-III+ instead of K6-2.

I've bought one specifically for my VLB boards. The problem is, it requires a BIOS upgrade or at least a BIOS that doesn't prevent the system from booting with a non-Pentium chip. The kit comes with a CD full of MR.BIOS and Unicore upgrades, but the only supported Opti chipset is Viper.

So as far as upgrade kits go, Powerleap/Madex would probably be better, because they're designed for user-installed CPUs. I've tried removing K6-2 from my Spectra and it's really a tight fit — still afraid I'll damage it.

Last edited by jheronimus on 2021-09-09, 16:35. Edited 1 time in total.

MR BIOS catalog
Unicore catalog

Reply 8 of 20, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Personally I only find these um (aberrations isn't a kind word is it?) "unique" boards interesting "as is" meaning running the P90 or so that they would have used when new. Pimping them out beyond "best of era" parts feels like putting dubs on Grandma's Hyundai Excel. Just amplifies the quirky and makes the steering harder in the supermarket parking lot. When I dig mine out I feel like I'd just be sticking a 4MB Virge, SB16, 4GB EIDE and a P120 in it, if I've got the "right" one, or the gold top P90 if not. (Weird multiplier arrangements on 1st gen socket 5, it's not clear that all boards support 2x on later P54)

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 9 of 20, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t
WJG6260 wrote on 2021-09-09, 15:56:

It’s funny you mention that. I was looking around and seem to find that these are the fastest Socket 5-compatible overdrives. Unfortunately, it seems like there are few out there. Even the Pentium OverDrive with MMX seems tougher to find than it should be; I find this confusing, as presumably many were made? It seems like there’s plenty of Socket 3 PODs available, but I’ve had one for years, and that’s not too helpful in this case haha.

The reason so many POD83s survived is because they were expensive and highly incompatible upgrades, and very few people actually bought them at the time of release. They sat around in warehouses in their retail packaging until they finally got listed on eBay at discount prices in the mid 2000s and snapped up by enthusiasts.
Most other types of overdrives were unboxed and used, so were more likely to go to the scrappers.

WJG6260 wrote on 2021-09-09, 15:56:

Out of curiosity, I haven’t seen the chipset on your Pentium VLB setup used elsewhere—was it designed for active backplane operation?

I haven't seen the OPTi EISA Pentium chipset used on other boards either. I've seen the original sales brochure, and I don't think it was designed for backplanes. I think it was just way too expensive to implement, so rarely used.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 10 of 20, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++
Anonymous Coward wrote on 2021-09-09, 17:11:

The reason so many POD83s survived is because they were expensive and highly incompatible upgrades, and very few people actually bought them at the time of release. They sat around in warehouses in their retail packaging until they finally got listed on eBay at discount prices in the mid 2000s and snapped up by enthusiasts.

Yup, intel ballsed up the overdrive specification they gave out a couple of years prior, so 5V only boards, probably installed with a DX, didn't support the POD, despite initially promising to, then by the time that spec got fixed, the boards that could run a POD could also natively run a DX4 and given 16 bit integer code of the time, and the pentium killer apps not really being a thing until 2 or 3 years later, DX4s didn't look like they were hurting much against the POD at 4 times the price. So if you had the older 486 board, it was put up with DX2 66 max or replace the board... if your board was AT format. So those that could use an AT board looked at the price of a newer 486 board plus POD and the price of a socket 5 board plus P100 and said, NO! to the POD, for which price in the UK I think you could get a P120 plus motherboard even. I think low price motherboard plus P75 was even half that of POD83 pricing.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 11 of 20, by jheronimus

User metadata
Rank Oldbie
Rank
Oldbie
BitWrangler wrote on 2021-09-09, 17:41:

So those that could use an AT board looked at the price of a newer 486 board plus POD and the price of a socket 5 board plus P100 and said, NO! to the POD, for which price in the UK I think you could get a P120 plus motherboard even. I think low price motherboard plus P75 was even half that of POD83 pricing.

Not to mention the 5x86 chips that came out a year prior to POD, I think those were much cheaper as well. Granted, the motherboard compatibility is as rare as with the PODs, if not rarer. And real life performance difference would be negligible as well.

MR BIOS catalog
Unicore catalog

Reply 12 of 20, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Yeah I got the cyrix at the time which was approx 2/3 the price of a P75 with the motherboard for each. However, finding since that the P75s overclock like stink, I would probably have stretched for the P75 had I known. (At the time it looked like same performance for 2/3 price, not potentially double performance for only 50% more)

edit: I think the memory situation played into the decision also, I already had 4x1MB 30 pin plus one 2MB 72pin which could be used in 486 board, whereas I would need to buy 2x4MB 72pin for minimum config of P75 board, prices hadn't come down from epoxy factory fire, and win95 release demand kept them pegged up.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 13 of 20, by cyclone3d

User metadata
Rank l33t++
Rank
l33t++

I was going to mention the Evergreen and similar upgrade interposers as well.

I actually have a couple of them that I was able to scrounge up but I am most definitely not getting rid of them.

There were a few stand-alone interposers on eBay a few months ago and I snatched up some of those as well. (Edit: I'm going to have to look at these... I think these may actually be the socket 3 interposers. / Edit)

With the 66Mhz fsb, the best you are going to get with any CPU (AMD only) is 396Mhz (66x6 via the 2x multiplier).

Yamaha modified setupds and drivers
Yamaha XG repository
YMF7x4 Guide
Aopen AW744L II SB-LINK

Reply 14 of 20, by WJG6260

User metadata
Rank Member
Rank
Member
jheronimus wrote on 2021-09-09, 16:03:

Thank you so much for the manual — this is simply an insane find! At a quick glance it doesn't offer much insight into those jumpers (I was hoping there's a VLB speed divider somewhere), but I'm sure it will give some useful info.

My pleasure! I'm glad to share it! I have been meaning to make a copy of it, and when you asked, it reminded me that that was on my to-do list. It really does not offer much to tell us what those myriad of "factory default" jumpers do, but perhaps there's someone out there who's experimented? I haven't found much. It seems that these boards were so unpopular and uncommon that perhaps we're some of the first to really tinker with them and explore.

I believe both boards are very picky about the RAM you use. I ran memtest on my Biostar with a set of FPM sticks that I used on a lot of my 486 boards with no issues and it simply gave me a ton of errors. So maybe I should run the same test on Supermicro and I could use tighter timings with a different set of RAM.

Funnily enough, I found my boards to be less picky than I expected. I used standard FPM SIMMs, although they might be more compatible by virtue of being HP-branded NEC SIMMs. I have a bunch in my Vectra 486/33n, and liked them enough that I figured I'd give them a go here. I actually haven't tried memtest though; that's an interesting call--I might check and report back on that. I wonder how the Northgate does with a WinChip2 in something more diagnostic and strenuous like memtest. I should see and report back.

I can't remember the specifics of using alternate BIOSes, but I remember that in general they were slower than the original Supermicro and somewhat less stable. I was having high hopes for the Biostar boards, because I've thought that Award boards didn't block the non-Pentiums, but it's not the case with Biostar, unfortunately.

Dang, this is not what I wanted to hear. I figured they'd be less stable, but slower too? That's not great. I guess I can live with the tricky SuperMicro BIOS and see how things go. The Biostar board might be a better option, just as my Northgate might similarly be more of a compelling pick, in such an instance. I guess I can really test and see once I replace the CMOS battery. I have one sitting on my desk, but haven't had time to solder it on, as I just started classes. Hopefully, this weekend or perhaps next week will provide more time haha.

Huh, guess I might have a dead WinChip after all, I should test it on a more common board.

I sure hope not! Please keep me posted! I know that, for some reason, the Northgate is weird when I configure it in specific ways. For example, it likes certain VLB cards, like my Trio32, and then sometimes it won't play nice, and I have to pop-in a Vision864, which it seems to work just fine with always. It must be something with that Phoenix BIOS. I'm not the biggest fan, as you can probably guess.

It even says the same on their website.

As to whether I have a later, "fixed" revision of the board, or if perhaps the 546 and 596 are interchangeable, I am not certain, but the fact remains that there is reason to suppose both could be true.

Interestingly both our boards are rev.3, but yours has these red elements (I don't know what they're called) while mine doesn't. Don't see any other difference in layout.

This is something I'm quite interested in. I did also notice that my SuperMicro board only has the OPTi Premium branding on the 597, but not on the 546. What a weird thing. Heck, even the board on Ultimate Hardware 2019 doesn't have the chipset configuration that mine does...

I wonder if the difference is more-or-less superficial, like the PCB revision differences on the Alaris SLC2/BL3 boards. I, for example, have an RYC/Alaris 486SLC2 that has a slightly different layout and slightly different colored components than some other forum members here do. I believe mine is slightly different from Anonymous Coward's, for example. But, performance is the same, and I figure that the same is probably true here. Although, that 546 might play a bigger role in that; or it might not, and I'm pontificating aimlessly haha

If I am not mistaken, few, if any of these boards were sold in retail, most of them were only available via small system builders. Again, according to Usenet, Supermicro didn't really provide user support — it's a miracle they even released a BIOS update or even acknowledged the board on their website. Then again, it seems to be one of their first products ever (the company started in 1993, a year before this board's time).

Makes sense. BitWrangler brings up a good point there too. These boards were not top-of-the-line, but more end-of-the-line. It's a shame, too. I happen to own a revision 1 of Anonymous Coward's NICE VL/EISA board (it's the same, but with a LIF socket), and it's like a distant cousin of this board, but, unlike the P54VL-PCI, it was just truly the highest end at the time. VLB and Pentium is a match made in hell, but it isn't completely awful if implemented well. I'd imagine that the Python/Cobra can perform at least in the vicinity of a 430LX/NX if configured properly. That said, the 430LX/NX are certainly going to have higher ceilings, even if their floor is approximate to the OPTi's maximum performance.

It's weird that you can't use PCI video (I didn't have that issue on either board). I think WaybackTech had the same exact problem with his TMC Socket 5 board, but he mentions he used a non-original BIOS. His board used a 571/572 chipset and acted very strangely in many regards and I think he had memory bandwidth that was even below 486 levels. So yes, it could be related to BIOS.

I gave this video a watch and found it interesting. He definitely has more issues with the BIOS than I do, but the Phoenix BIOS on that board just does not do it wonders. I'd be interested to try another, but my concern is about the integrated I/O being non-functional. Hilariously, that I/O is VLB-based, although I suppose that makes sense, since the PCI on these boards is bridged to the VLB anyway. Perhaps the direct VLB I/O is marginally faster than a bridged PCI->VLB I/O setup? Either way, I digress.

I don't think any regular MMX chip is rated for 3.3V. I haven't tried Win95 on Supermicro at all — only on Biostar, so can't comment there. But generally compared to DOS/Win3.11, Win9x is a lot more sensitive once you start picking tighter timings.

Out of curiosity, was the Win9x experience decent on the Biostar? I used to run Windows 98SE on a P166 and an AOPEN AP5C/P 430FX board, and it was more than usable. That was a main machine of mine for a while, and a family friend's before that.

That would be really cool. I've been looking for an image for a very long time in many places 😀

Hopefully we can find it sometime! It'd be a shame to be lost to history. I have been searching (and I believe Anonymous Coward has as well) for the MR-BIOS for the OPTi 295 chipset too. It would be great fortune if they happened to be on the same disc, and someone found that disc.

Off-topic here, but on the topic of nonexistent drivers and whatnot, I recently worked on a Paradise Accelerator Pro (IIT AGX-014 card), and getting that thing running was a nightmare. I had to use a VESA TSR for the Orchid Celsius in DOS, and Windows 95, as that had built-in drivers. Side note: These AGX cards are really weird. I would not recommend one for Windows acceleration either. Perhaps they were nice in Win3.1 with drivers...

Hopefully, our experience here fares differently and we can find these lost MR-BIOSes.

cyclone3d, I think you're right about that. Unfortunately, 66x1.5 is going to be the hard limit here, unless I were to bridge the BF0/BF1 pins and hard-mod the board or CPU for such an application. There's also the WinChip option, which reinterprets 1.5X as 4X.

As for the interposers, I saw a few on eBay for 486s, and grabbed one a while ago. There's an overpriced Socket 5/7 one out there that the seller wants an exorbitant amount for, so I'm learning towards finding an MMX OverDrive for this purpose. That, or the WinChip2 at 50x4.

Speaking of, anyone know how a 200MHz WinChip2 would compare against a Pentium MMX 166? I'd imagine they'd be close, no? The MMX 180 is probably the more intriguing OverDrive chip, in such a case.

Anonymous Coward wrote on 2021-09-09, 17:11:

The reason so many POD83s survived is because they were expensive and highly incompatible upgrades, and very few people actually bought them at the time of release. They sat around in warehouses in their retail packaging until they finally got listed on eBay at discount prices in the mid 2000s and snapped up by enthusiasts. Most other types of overdrives were unboxed and used, so were more likely to go to the scrappers.

Makes sense to me. I bought mine about 15 years ago on Amazon, and it was NIB. I've since used it a bunch of times, and it's been great. It was originally intended for my aforementioned Vectra 486/33n, which is supposed to be compatible. Well, it wasn't--until I found the updated BIOS sometime a few years back. It works fine now, but for years it was a nice testing chip and that's about it.

I haven't seen the OPTi EISA Pentium chipset used on other boards either. I've seen the original sales brochure, and I don't think it was designed for backplanes. I think it was just way too expensive to implement, so rarely used.

Huh, go figure. I haven't seen too many EISA Pentium boards anyway. It seems like there's more 486 ones out there, and the SiS chipset on the NICE Super VL/EISA is more than fast enough for some fun EISA action. I did see a dual Socket 5 EISA board a while back on eBay, but the price hurt to look at, and EISA on a Pentium isn't as compelling for some reason, at least not to myself.

-Live Long and Prosper-

Feel free to check out my YouTube and Twitter!

Reply 17 of 20, by chublord

User metadata
Rank Member
Rank
Member

I had one of those Supermicro boards before and enjoyed it. Had a SCSI setup and a Matrox PCI card, I foolishly gave it away.

IBM Valuepoint 486 DX4-100, Opti 802G, 50 MHz FSB, Voodoo1+S3 864, Quantum Fireball EX 4.0 GB, Seagate Medalist 1.6 GB, 128 MB FPM, 256k L2

Reply 18 of 20, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Huh, go figure. I haven't seen too many EISA Pentium boards anyway. It seems like there's more 486 ones out there, and the SiS chipset on the NICE Super VL/EISA is more than fast enough for some fun EISA action. I did see a dual Socket 5 EISA board a while back on eBay, but the price hurt to look at, and EISA on a Pentium isn't as compelling for some reason, at least not to myself.

EISA wasn't interesting on Pentiums unless you needed to run a LOT of network cards, or as a home user if you didn't have PCI slots (very rare). I think compaq even made a system that only had EISA. Come to think of it, I am pretty sure there are P5 boards with only ISA slots if you can believe that!
I just picked up a dual Socket5 from DTK with neptune chipset for $50 with free shipping on eBay. It seems you can still get them cheap sometimes.Though, It'll be quite a while before I can test it, so hopefully it works! To bad it doesn't have VLB slots!

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 19 of 20, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

I could imagine someone wanting one in 95 or 96 to run the caching scsi controller array for the server that's beating a dx2 66 into the ground right now, but the disk performance still adequate.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.