These were the main boards we used for 5x86 and late-model 486 CPUs. Most Red Hill systems sold in the mid-nineties were built around one or another of them. They are no longer manufactured, of course, and it became rare to even use one for a repair job towards the end of the decade, but we would still see them coming in from time to time, perhaps for a belated upgrade or a small repair, but quite often for something unrelated. It is surprising how many people found that an old DX/2, DX/4 or 5x86 could still do wordprocessing and perhaps a little email, which is all they needed. These old boards soldiered on for a long time in business systems too, perhaps as print servers or controlling industrial machinery.

FIC 486 PVT

Always one of our favourite VESA 486 boards. Most of our 486DX systems used either one of these or a QDI OPTi 895 (below). The FIC board was slightly faster than the QDI — and main board speed used to matter in those days — but was inclined to be fussy about VESA timing, and so took rather more care to set up. The superb Award BIOS was ahead of its time; it had a very similar look and feel to the ubiquitous Award BIOS we would all take for granted a few years later on when Pentium IIs and K6-IIIs were standard fare.

The PVT had one very odd quirk: if you loaded a standard Microsoft mouse driver in your autoexec.bat but you didn't actually have a mouse plugged in, the system would almost always refuse to boot. We never did work out why.

  • CPU support: Any 486, very late revisions support 5x86 too.
  • Speed: 25, 33, or 40MHz.
  • Slots: 2 VESA, 6 ISA
  • I/O: None. LBA support only fair at first.
  • RAM: 4 72-pin Fast Page.
  • Cache: Socketed, 256k asynch standard, 512k optional.
  • Chipset: VIA Pluto, Award BIOS.
  • Best with: 486DX/66, DX/100.
  • Status: Legacy.
QDI OPTi 985

QDI OPTi 895

Overall, our favourite 486 VESA board. Very reliable, and remarkably stable with difficult CPUs like the 486DX/2-80.

These gave us a wonderful run as our main 486 board for over 12 months — which in the computer industry is a long time indeed! But all good things come to an end, and they finally stopped making the 895 late in '96, though we sold so many that we were trading them back in for years after that, and still used to see one now and then until early in the new century.

Our only complaint about the 895 was the poor jumper layout: messy and very hard to memorise. Oh, and that terrible AMI mouse-based BIOS interface. (More on AMI's then-new BIOS shortly.)

Illustration: the QDI OPTi boards came in a number of versions; you can tell the approximate age of the board by the RAM arrangements: early ones had eight 30-pin sockets; the majority, like the one illustrated, had four 30-pin and two 72-pin sockets; and the final version in VESA's dying days had no 30-pin sockets but room for four 72-pin SIMMs.

By the way, we still get masses of email asking how to set the jumpers on these boards, usually from people wanting to fit a CPU for which they were never designed. (Yes, still — and it is now April 2004.) The QDI OPTi895 came in a number of different versions, each with different settings, and it was only one of dozens, maybe hundreds of different boards based on the 895 chipset. So the chances that we can guess the (possibly nonexistent) correct jumper settings for your old OPTi 486 board without actually seeing it in the workshop are very slim indeed.

  • CPU support: Any 486, later revisions support 5x86 too.
  • Speed: 25, 33, or 40MHz.
  • Slots: 3 VESA, 6 ISA
  • I/O: None. LBA support only fair at first.
  • RAM: 8 30-pin, 4 30-pin and 2 72-pin, or 4 72-pin, up to 64MB.
  • Cache: Socketed, 256k asynch standard, 512k optional.
  • Chipset: OPTi 895, AMI 'Windows' BIOS.
  • Best with: 486DX/66, DX/80, DX/100.
  • Status: Legacy.

BIOS

For years AMI dominated the BIOS market. Most late-model 286 boards and nearly all 386 and 486 boards used AMI's excellent Colour BIOS. AMI's interface was simple and very easy to use.

Then, around the time that the 486DX/4 became popular, AMI switched to a new graphical, mouse-based Windows BIOS, which was supposed to be more user-friendly. It was terrible: the interface was clumsy, slow, counter-intuitive, and a real pain in the workshop where time is money and efficiency is everything.

Meanwhile, the once-popular Award BIOS had become a bit of a rarity — so much so that Award changed the traditional dark blue interface to look and feel more like the near-universal AMI Colour BIOS. No-one took much notice — until AMI introduced the idiotic new Windows BIOS, which actually looked more like an old Commodore-64. We assume that this is why Award surged ahead to dominate the BIOS market in the late 1990s. Up until just before the turn of the century, an AMI BIOS in a new system became quite a rarity.

Illustrations: On the left, the excellent AMI Color BIOS interface that drove all competitors into niches from about 1990 on. The particular one illustrated is from a 386DX-40 board.

On the right, the clumsy, childish and infuriatingly slow AMI Windows BIOS that arrived around 1994, to universal dislike. Thankfully, the new Award BIOS (not illustrated) was, like the old AMI one, very usable and it soon took over as the standard — driving AMI almost out of the market, which served them right.

Octek Hippo DCA2

A board we never sold new, though not for any want of trying on the Octek sales team's part. Around 1994 or 1995 Octek invented a concept they called Dynamic Cache Arcitecture which (they claimed) made Octek boards faster than any other board on the planet. And yet — notice the complete absence of cache RAM!

Rather than supply a mere 256k of cache RAM, Octek claimed that their boards 'adapted themselves' to cache the entire system RAM! And since no-one else had Dynamic Cache Arcitecture, only Octek boards could deliver.

For a year or two, DCA boards were everywhere: Octek, long just one of many board makers, was now highly visible. Better yet, the DCA boards were not a great deal dearer than standard boards, and came bundled with a stick of their special DynamiCache RAM.

These were extraordinary claims, on the face of things too good to be true. The ability of the Octek sales team to make sweeping claims but simulteanously dodge awkward questions became legendary.

What the Octek boards actually turned out to be, so far as we could ever discover, were perfetly ordinary 486 main boards with no cache and a modified BIOS. The board illustrated has 8MB of Octek's DynamiCache fitted, and in that configuration it would have performed very well — as well or better than a standard board with 256k of standard cache. The difficulty began when you wanted to add more RAM, for although the Octek boards could use any normal RAM, they only delivered their promised performance when equipped with Octek's extraordinary 15ns stuff. (Was it really 15ns main RAM, with no hidden catches? If so, why did Intel not buy masses of it for their new cost-no-object Pentium 66 boards? And why do Octek no longer exist?)

  • CPU support: 486.
  • Speed: 33 MHz.
  • Slots: 3 VESA, 3 ISA, 2 8-bit
  • I/O: None.
  • RAM: 4 72-pin, fast page or DCA.
  • Cache: None.
  • Chipset: OPTi 802, Mr BIOS.
  • Best with: 486DX/66, DX/100.
  • Date: Around 1995.

QDI P4U885P3

One of the first 486 boards to offer a PCI bus instead of the till-then universal VESA, and a very big seller for a short while. These were never as solid as the VESA boards of the same period, or the more mature PCI boards which followed. Early models (like the one in the illustration) had twin 32-bit hard drive controllers on board, later revisions had a full I/O section, including 16550 UARTs.

The Phoenix BIOS was rather odd and you had to get to know it. Even back then, Phoenix had a long-established tradition of writing dreadful user interfaces, and this one was no exception: end users often had a lot of difficulty understanding it.

Logical Block Addressing (LBA), which is needed for hard drives over 528MB, was new when this board came out, and — you guessed it — didn't work as well as it did on later boards. As a rule-of-thumb, first generation LBA works well but you can't change to a different brand of motherboard unless you re-partition the hard drive. Other boards from the same time were similar — the OPTi 895 and FIC PVT are examples. We assume that it was related not to the board as such but to the LBA BIOS extensions.

  • CPU support: Any 486
  • Speed: 25, 33, or 40MHz.
  • Slots: 3 PCI, 4 ISA
  • I/O (early model): 2 32-bit HDC only.
  • I/O (later model): 2 32-bit HDC, FDC, 2 high-speed serial, one parallel. LBA support only fair.
  • RAM: 4 72-pin Fast Page.
  • Cache: Socketed, 256k asynch standard, 512k optional.
  • Chipset: UMC, Phoenix BIOS.
  • Best with: 486DX/66, DX/100.
  • Status: Legacy.

FIC PIO-2

Forerunner of the wonderful PIO-3, and another great 486 board. Quite large by 486 standards, but very easy to work on and beautifully documented. (A traditional FIC strength.) One of the first really reliable 486 PCI motherboards.

PCI was developed by a consortium led by Intel, which seems to have been much more interested in getting it to work properly in Pentium-based boards (and even those were dreadful in the early days). Once the technology matured, PCI became quite fuss-free and caused very few problems. The assignment of interrupts to it remained problematic for many years, but this was a Plug and Play issue and we can't really blame the bus for that.

  • CPU support: Any 486
  • Speed: 25, 33, or 40MHz.
  • Slots: 4 PCI, 4 ISA
  • I/O: 2 32-bit HDC, FDC, 2 serial, one parallel, LBA.
  • RAM: 4 72-pin Fast Page, up to 64MB.
  • Cache: Socketed, 256k asynch standard, 512k optional.
  • Chipset: VIA Pluto, Award BIOS.
  • Best with: 486DX/66, DX/100.
  • Manufactured: September 1995.
  • Status: Legacy.
FIC VIP I/O

FIC VIP-I/O

The transition from VESA to PCI was a rocky one for many computer owners. If you had a VESA system and you had bought it a fast video card, say a Number 9 or a Diamond Stealth, you faced a very expensive upgrade path. If you bought a new PCI motherboard, you had to shell out an extra $300 or so to replace the video card as well. On the other hand, if you stayed with VESA you were locking yourself into a dying technology.

This was the purpose of the VIP-I/O: it cost about $100 more than a VESA board and $50 more than a PCI board, but you could mix and match.

In our office server we used a VIP-I/O to extend the working life of our QLogic VESA Fast SCSI host adaptor (which we were reluctant to retire as it owed us $500) but still upgrade to a superb new Number 9 2MB PCI video card. It worked too. We got another year or two of service out of the QLogic card before the SCSI rig gave way to IDE for a while; later the VIP-I/O came out in favour of a PA-2005, a PA-2007, an Iwill P55XPlus, and a succession on MVP3-based boards, but the old Number 9 was still in our main file server till the turn of the century. The VIP-I/O had done its job: preserved our investment in the SCSI card, and still allowed us to move to PCI for video. Along the way, it probably saved us five hundred dollars.

Because of the cost, we only sold about ten VIP-I/Os, mostly to people with similar needs to our own, but they were all still in service until late in the decade.

  • CPU support: Any 486.
  • Speed: 25, 33, or 40MHz.
  • Slots: 2 VESA, 3 PCI, 4 ISA
  • I/O: 2 32-bit HDC, FDC, 2 serial, one parallel, LBA.
  • RAM: 4 72-pin Fast Page.
  • Cache: Socketed, 256k asynch standard, 512k optional.
  • Chipset: VIA Pluto, Award BIOS.
  • Best with: 486DX/4-100.
  • Status: Legacy.