VOGONS


First post, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t

I have been refurbishing a system I just bought, with the intent of using the really nice desktop AT case to house a purpose-built test system for VLB and ISA cards from the 486 era.

Problem is, I am having trouble deciding on a motherboard. I have a decent collection of 486 boards but a lot of them still need batteries. Of the ones that have batteries and hopefully work (still need to test most of them) I have the following:

PC Chips M912 v1.7 (UMC UM8498F chipset, looks like real cache)
Acer Vi15G (SIS 85C417, cache, SMD tantalum caps... looks like a fancy board for the time)
PC Chips M919 v3.2 (UM8881F chipset, PCI + VLB, onboard IDE, but has fake cache unless I can ever find an M919 cache card for it)
ECS AL486 (ALI M1429G chipset, real cache, tested and working last time I tried it)
DFI G486VPA (VIA VT82C496G chipset, PCI + VLB, later board from 1995-1996, onboard IDE, cache)

My goal is maximum compatibility, since it will be used for testing cards. I want the board to be "the problem" as rarely as possible when testing cards. I have done a ton of research but I really don't have a lot of hands-on experience with 486 systems so I don't know if there is a big difference in compatibility or conflicts with any of these chipsets or boards, particularly when dealing with VLB video cards. I will also try to equip the board with a decently fast processor so that if I choose to do some benchmarking or comparisons it won't be too much of a bottleneck. At least a DX4 100Mhz, but possibly something faster like an AMD 5x86 133 or PODP5V 83Mhz.

I am sort of leaning toward the DFI G486VPA because for a testing machine I think the onboard drive controllers will be a bit more convenient, and if there's any possibility that they are PCI based I'm sure that will help to avoid conflicts with VLB cards I may be testing... yet, I can't find any recommendations for VLB boards with Via chipsets, so I am worried that there's a reason for that.

I'm not trying to break benchmark records or maximize performance in games... I mainly want stability and flexibility for testing.

Any input you guys can offer would be greatly appreciated. 😀

Now for some blitting from the back buffer.

Reply 1 of 4, by PC@LIVE

User metadata
Rank Oldbie
Rank
Oldbie
Ozzuneoj wrote on 2025-03-21, 05:39:
I have been refurbishing a system I just bought, with the intent of using the really nice desktop AT case to house a purpose-bui […]
Show full quote

I have been refurbishing a system I just bought, with the intent of using the really nice desktop AT case to house a purpose-built test system for VLB and ISA cards from the 486 era.

Problem is, I am having trouble deciding on a motherboard. I have a decent collection of 486 boards but a lot of them still need batteries. Of the ones that have batteries and hopefully work (still need to test most of them) I have the following:

PC Chips M912 v1.7 (UMC UM8498F chipset, looks like real cache)
Acer Vi15G (SIS 85C417, cache, SMD tantalum caps... looks like a fancy board for the time)
PC Chips M919 v3.2 (UM8881F chipset, PCI + VLB, onboard IDE, but has fake cache unless I can ever find an M919 cache card for it)
ECS AL486 (ALI M1429G chipset, real cache, tested and working last time I tried it)
DFI G486VPA (VIA VT82C496G chipset, PCI + VLB, later board from 1995-1996, onboard IDE, cache)

My goal is maximum compatibility, since it will be used for testing cards. I want the board to be "the problem" as rarely as possible when testing cards. I have done a ton of research but I really don't have a lot of hands-on experience with 486 systems so I don't know if there is a big difference in compatibility or conflicts with any of these chipsets or boards, particularly when dealing with VLB video cards. I will also try to equip the board with a decently fast processor so that if I choose to do some benchmarking or comparisons it won't be too much of a bottleneck. At least a DX4 100Mhz, but possibly something faster like an AMD 5x86 133 or PODP5V 83Mhz.

I am sort of leaning toward the DFI G486VPA because for a testing machine I think the onboard drive controllers will be a bit more convenient, and if there's any possibility that they are PCI based I'm sure that will help to avoid conflicts with VLB cards I may be testing... yet, I can't find any recommendations for VLB boards with Via chipsets, so I am worried that there's a reason for that.

I'm not trying to break benchmark records or maximize performance in games... I mainly want stability and flexibility for testing.

Any input you guys can offer would be greatly appreciated. 😀

Ciao
In my collection I have a certain number of 486 Motherboards, most are VLB 486DX2-66, but I have other working ones, I would start by saying that I don't have a 486 PCI-VLB card, it's already this in my opinion, it's an ideal card for a 486, so the other cards only VLB, I would keep them in second choice, but I wouldn't discard them completely.
Among those on your list, if I read that right, there are two motherboards with PCI-VLB, but which is better between the two, a comparison should be made, for example with Phil's benches (or other similar ones), you can see which of the two makes the best scores, in theory it should be discounted, the DFI does better (?), but the PCChips with false cache, could if you update the cache to real, get more or less equal scores, and at that point, I would tell you to choose the card with more ISA PCI slots (usually the VLB is 1), or if you need more VLB slots, because you have to try video cards and controllers, you should go on the ISA-VLB cards, so you should redo the benches, and see which one does the best.
In the past I made a comparison on my four (then) 486DX2-66 VLB, I discovered that two have 0KB of cache, and they are slow, one of these has the real cache, unfortunately it does not work, I will try a repair as soon as possible, I had tried years ago, there is in the first pages of "Test...PC@LIVE ... Motherboard".
As a CPU what to recommend, it depends on which card you choose, those with PCI I see them more suitable for 486DX4-100 or higher, while the VLBs more towards 486DX4-100 or lower, personally I will upgrade from 486DX2-66 to 486DX2-80 on a VLB, without overclocking, and I would do a second upgrade on a 486 ISA (not VLB or PCI), always with a DX2-80, but in 5V version.

AMD 286-16 287-10 4MB HD 45MB VGA 256KB
AMD 386DX-40 Intel 387 8MB HD 81MB VGA 256KB
Cyrix 486DLC-40 IIT387-40 8MB VGA 512KB
AMD 5X86-133 16MB VGA VLB CL5428 2MB and many others
AMD K62+ 550 SOYO 5EMA+ and many others
AST Pentium Pro 200 MHz L2 256KB

Reply 2 of 4, by Intel486dx33

User metadata
Rank l33t++
Rank
l33t++

From my Experience.

1) 486dx2-66 or Faster. You can always Slow down the CPU Speed by Disabling Motherboard cache or CPU cache by using the program “SetMul” ( Set multiplier )

2) Allot of VLB controller drivers have a limit of 410mb hard drives in Win3x
So use Win95b and Use a Compact Flash card ( CF ) for Best Performance and to get around the Driver limitations.

4) The Very First Upgrade I would do is to get a CF card as a hard drive
I like the one by “Startech” you can mount it anywhere. Good Build Quality.
Don’t buy a Cheap one from China. They break easily.

5) Memory
8mb of Memory should be the Minimum for DOS and Win95
16mb is Very Good.
32 in Maximum and will only be accessible in Win3x or Win95
Load Device driver HIGH in Configuration files

6 ) Win95b is the Easiest and Best OS for old ISA cards and 486 hardware.

Reply 3 of 4, by douglar

User metadata
Rank l33t
Rank
l33t

I have a PC Chips 919 and I like it. Runs Am486 @ 160 no problem. I got a cache stick here:
Lets make new M919 Cache sticks?
But if you can run low wait state EDO, programs that use a memory footprint bigger than your cache (Doom, Quake) might perform better without the cache overhead.

VT82C496G? Does that one run PCI off a 505 PCI bridge chip like the VT82C486? I have one of those. Performance and compatibility suffers.

I have a dead ALI M1429G board that probably has a bad tantalum cap. Fixing it is on my todo list.

Reply 4 of 4, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t
PC@LIVE wrote on 2025-03-21, 07:33:
Ciao In my collection I have a certain number of 486 Motherboards, most are VLB 486DX2-66, but I have other working ones, I woul […]
Show full quote

Ciao
In my collection I have a certain number of 486 Motherboards, most are VLB 486DX2-66, but I have other working ones, I would start by saying that I don't have a 486 PCI-VLB card, it's already this in my opinion, it's an ideal card for a 486, so the other cards only VLB, I would keep them in second choice, but I wouldn't discard them completely.
Among those on your list, if I read that right, there are two motherboards with PCI-VLB, but which is better between the two, a comparison should be made, for example with Phil's benches (or other similar ones), you can see which of the two makes the best scores, in theory it should be discounted, the DFI does better (?), but the PCChips with false cache, could if you update the cache to real, get more or less equal scores, and at that point, I would tell you to choose the card with more ISA PCI slots (usually the VLB is 1), or if you need more VLB slots, because you have to try video cards and controllers, you should go on the ISA-VLB cards, so you should redo the benches, and see which one does the best.
In the past I made a comparison on my four (then) 486DX2-66 VLB, I discovered that two have 0KB of cache, and they are slow, one of these has the real cache, unfortunately it does not work, I will try a repair as soon as possible, I had tried years ago, there is in the first pages of "Test...PC@LIVE ... Motherboard".
As a CPU what to recommend, it depends on which card you choose, those with PCI I see them more suitable for 486DX4-100 or higher, while the VLBs more towards 486DX4-100 or lower, personally I will upgrade from 486DX2-66 to 486DX2-80 on a VLB, without overclocking, and I would do a second upgrade on a 486 ISA (not VLB or PCI), always with a DX2-80, but in 5V version.

Thank you for the input.

Intel486dx33 wrote on 2025-03-21, 09:38:

From my Experience. ...

Thanks, but did you mean to post this in another thread? This doesn't seem to have anything to do with my post and doesn't even mention anything about motherboards.

douglar wrote on 2025-03-21, 14:58:
I have a PC Chips 919 and I like it. Runs Am486 @ 160 no problem. I got a cache stick here: Lets make new M919 Cache sticks? […]
Show full quote

I have a PC Chips 919 and I like it. Runs Am486 @ 160 no problem. I got a cache stick here:
Lets make new M919 Cache sticks?
But if you can run low wait state EDO, programs that use a memory footprint bigger than your cache (Doom, Quake) might perform better without the cache overhead.

VT82C496G? Does that one run PCI off a 505 PCI bridge chip like the VT82C486? I have one of those. Performance and compatibility suffers.

I have a dead ALI M1429G board that probably has a bad tantalum cap. Fixing it is on my todo list.

I considered building one of those DIY cache sticks a while back but judging from a lot of the posts there I see a lot of potential for failure and wasted time\money. I don't have much experience with SMD multi-leg chip soldering and basically none with SOJ chips. I still keep my eyes open for an original M919 cache stick when I think of it... but I've had this board for around 7 years at this point and haven't had an opportunity to grab one. 😔

With regard to the low wait state EDO, I'm not really familiar with wait states at all. How does that work and how do I verify that I have that kind of memory? And if I'm just using the system for testing basic software and games under DOS and Windows (98 or 95 probably), does this offset enough of the performance penalty of running with no external cache to make the system still viable?

As for the DFI board with the VIA chipset, it does have the VT82C505. I see on TRW it says that this is a VLB to PCI bridge... so they just have PCI running off of the VL Bus? Ouch. That doesn't sound ideal at all. It's too bad, because the board looks awesome otherwise.

If there was JUST a minor\moderate performance impact I would probably just use it anyway... but if it makes certain VLB cards unhappy or makes the system unstable or unreliable, then it won't be good for a tester. If anyone has more info about the specifics of the VIA VLB to PCI bridges it would be much appreciated. I won't be using this to test PCI cards... just VLB and ISA. PCI would be nice if it can provide storage and I\O that don't interfere with whatever I'm trying to test, but if this bridge setup just adds more complication then that's too bad. 🙁

Now for some blitting from the back buffer.