1998 was a year of two distinct halves and quite dramatic transitions. In January we were still waiting for the 6x86MX-233 to arrive but by December we were well into the era of the era of the K6-2, the Voodo III, and the overclocked Celeron 300A. Two completely different worlds.

Nothing much went as expected in 1998, there were surprises everywhere and it is no wonder that it ended up being a year of major shortages. To begin with there was a noticeable trend to Slot 1 boards developing, but the arrival of the K6-2 and the non-event of the Intel Celeron reversed that. The original Celeron was such a poor performer that the expected surge in Slot 1 board sales did not eventuate. Intel made matters worse by restricting supply of the only two really salable Pentium II parts, the 266 and the 300, and their much improved Celeron-A came too late to make much difference.

GA686BX detail

The mainboard manufacturers were all wrong-footed by events. They had not expected the strong trend towards Super 7 and the K6-2/300, as opposed to the sluggish and expensive Slot 1/Celeron combination, and second, underestimated the continuing demand for entry-level products like the C6 and the 6x86MX.

In their sudden scramble to get Super 7 products out many manufacturers abandoned 66 and 75MHz Socket 7 too soon and late 1998 and early 1999 saw a dramatic shortage of entry-level and lower midrange motherboards. (66MHz Socket 7 still remained strong in early 1999 and lasted till about mid- winter of that year.)

Here at Red Hill, this left us with the excellent VA-503+ almost unchallenged as our high-end board of choice, but the low-end and midrange was in a state of flux. Our entry-level mainstays for the previous six or twelve months, the FIC VA-502 and the ASUS SP97-V, had slipped into end of life by the second half of 1998, and our favourite midrange boards, the Iwill P55XPlus and the Shuttle HOT-569 disappeared too. (A premature end for these fine boards, we felt.) For the upper midrange, we were torn between the Shuttle HOT-591T and the Epox EP-58MVP3C-M.

With our favourite entry-level Socket 7 boards all going end of life, we went back to QDI and tried the Advance 2 for a while but eventually switched to using the ubiquitous VA-503+ for almost everything. We had expected Slot 1 boards to take off in a big way but the K6-2/300 stopped that trend in its tracks. Intel priced the Pentium II out of the market, we couldn't give the Celeron away and — to our surprise — the Celeron-A didn't sell in any numbers either. In short, Slot 1 boards remained a bit of a rarity all year. (The vastly improved Socket 370 Celerons, by the way, arrived in early 1999 and were much more successful.)

FIC VA-502

FIC VA-502

Based on the same VIA Apollo VP-X chipset as the wonderful old PA-2005, these were our standard entry-level board for a long, long time (replacing both the PA-2005 and the Gigabyte 586s). They had all the important modern features like ATA-33, SDRAM, and 75MHz bus speed, performance was good, and reliability up to FIC's usual high standard. (Though not as good as the incomparable PA-2005.)

If you ran a 6x86MX-200, the VA-502 was simply the best board you could buy at any price. Those two made a magic combination. On the test bench with an MX-200, the 502 out-performed boards costing nearly twice as much, and on a VA-502 the 6x86MX-200 comfortably beat not just its direct competitors, the Pentium MMX and K6 Classic 200s, but the 233 MMX as well.

  • CPU support: 6x86, 6x86MX to 200, C6, K5, K6, P54C, P55C.
  • Speed: 50, 55, 60, 66 and 75MHz.
  • Slots: 4 PCI, 3 ISA
  • RAM: 2 168-pin SDRAM, 4 72-pin FPM, EDO or BEDO, up to 512MB.
  • In practice: EDO to 128MB, SDRAM to 64MB. 16Mb SDRAM chips only.
  • Cache: Surface mount, 256k or 512k pipeline burst.
  • Chipset: VIA Apollo 585VP, Award BIOS.
  • Best With: 6x86MX-166, 6x86MX-200.
  • Status: Legacy.

Single-sided and double-sided RAM

RAM slots from QDI Explorer 2

Generally speaking, boards of this era could accept any of the readily available varieties of 72-pin EDO or fast page RAM, so long as you fitted it in pairs. Common combinations were 16MB with a pair of 8MB EDO SIMMS, 32MB with two pairs of 8MB or one pair of 16MB, and 64MB with two pairs of 16MB or one pair of the rather unusual 32MB EDO SIMMS.

64MB EDO modules were available in theory but incredibly expensive and no-one could afford them. They may have found a home in servers or large CAD workstations but we certainly never sold any, and don't recall seeing one to this day.

The ability of boards to accept SDRAM varied quite a bit. Some of the oldest ones had no SDRAM slots at all. Some had only SDRAM support, the vast majority could work with either type — though very few boards could take both types at once the way a VA-503 could.

However, the vast majority of board up until about this time could only take SDRAM which used 16Mb chips. Note the lower-case 'b' to indicate mega bits, not mega bytes. Most 32MB DIMMs manufactured were double-sided — essentially a pair of 16MB DIMMs mounted back-to-back on a single circuit board — and each side had eight 16Mb chips to make 16MB, or 32MB in total.

Most 64MB DIMMs were single-sided, and used eight 64Mb chips. And this is where the problems start, for most older mainboards do not understand 64Mbit chips and see a standard 64MB DIMM as eight 16Mbit chips. Similarly, they see a typical double-sided 128MB DIMM as sixteen 16Mb chips and report only 32MB.

Elite P5SJ-B

Very similar to the ASUS SP97-V on a previous page, as it was based on the same SiS 5598 chipset. It was fractionally cheaper, slightly thinner (thick boards are usually the best), and hadn't had as much attention to detail in the layout and mounting requirements.

On the other hand, it supported SDRAM, which the SP97-V didn't, and SDRAM support was becoming important by this time. We had intended to replace the 97-V with these, but they went out of production before we had used more than a score of them. Indeed, "no longer manufactured" became an all too common problem about this time, as motherboard manufacturers scrambled to get out of 66MHz Socket 7 much too soon: demand for products like these remained strong for a good six months after all but a very few manufacturers had stopped supplying them.

(It's a tough industry. We could mention a score of other products which ended up rotting in warehouses because their makers over estimated market life — and we think retailing is hard!)

  • CPU support: 6x86, 6x86MX to 200, C6, K5, K6, P54C, P55C.
  • Speed: 50 to 75MHz.
  • Slots: 3 PCI, 3 ISA
  • RAM: 2 72-pin FPM or EDO, up to 256MB.
  • Cache: Surface mount, 512k pipeline burst.
  • Chipset: SiS 5598, Award BIOS.
  • Status: Legacy.

Iwill P55XPlus

We first saw these on the IBM Microelectronics website. The P55X+ was the board IBM used for benchmarking the brand new 83MHz bus 6x86MX-333, and as we had discovered with the MX-266, 83MHz was anything but easy to get right, so we thought we'd have a look at the Iwill product.

→ Illustration: an odd couple. By the time the Iwill P55X+ hit the market, 166MHz CPUs were long since finished as new products. It is a safe bet that the system originally shipped with a different board which needed to be replaced a year or two later. In fact it is probably on its third main board, as the 1.08GB hard drive suggests that it started life as a 486.

When they arrived, the Iwills boards were solidly constructed, beautifully packaged, and had the clearest, simplest jumper layout of any motherboard we had ever seen. (There is an illustration below.) Perhaps this is a small matter when compared to the big issues of price, performance, and reliability, but it is a small matter that very few mainboard manufacturers ever seem to get right. It is just so easy to think of a simple, practical layout and stick to it. The fact that some manufacturers, notably Iwill and Epox, do it as routine proves that it can't be all that difficult an art to master. Later on, we were to discover that this pleasing and practical simplicity was just part of the normal Iwill house style.

The chipset was the well-regarded ALI Aladdin 4+ — this was long before TNT cards arrived to break Ali chipsets, remember — and performance was excellent: right up there with the PA-2007 and VA-503+. We were even more impressed with the P55X+ in actual daily use. It was the only board we recommended for the often-difficult 6x86MX-233/75 at one time, and was as solid as a rock with the 83MHz bus 6x86MX-266. Even Super 7 boards struggled with the MX-266.


→ How to get jumpers right: the simple, obvious layout of the P55XPlus should have been an example to all manufacturers. It's London to a brick that this layout saved Iwill a considerable sum of money: a great many perfectly functional boards get returned for warranty service simply because the retailer has made a mistake with the jumpers: these were practically impossible to strap up incorrectly.

With only three PCI slots on the P55XPlus, it could be a struggle finding room for the I/O cables in some cases, but at least there was room for a long PCI card. The documentation was already out of date, but this was and still is a common sin — motherboards are almost always a step or two in front of the printed manual. Besides, these were so easy to set up that the manual was unnecessary. This too is the way a board should be. When the Iwills finished up we missed them.


  • CPU support: 6x86, 6x86MX, C6, K5, K6, K6-2, P54C, P55C.
  • Speed: 66, 75 and 83MHz.
  • Slots: 3 PCI, 5 ISA
  • RAM: 4 72-pin FPM or EDO and 2 168-pin SDRAM, up to 512MB.
  • Cache: Surface mount, 512k pipeline burst.
  • Chipset: ALI Aladdin 4+, Award BIOS.
  • Best With: 6x86MX-233 & 266, K6-233 and 266, K6-2/266
  • Status: Legacy.
ASUS P2L97 illustration

ASUS P2L97

Intel's long-awaited LX chipset was a vast improvement over their first P6 product, the FX. It took a lot longer than it should have to arrive, but was well worth the wait. Vastly faster than the FX, it added support for ATA-33 hard drives, SDRAM and AGP video.

The first two were fast becoming essential by this time, but as usual with all-new interfaces, the AGP video was very troublesome at first. Most buyers stuck with PCI for another year or so. In 1998 AGP was very prone to comparability and configuration problems: by the following year it was a tricky but reasonably sensible alternative, and by 2000 it was fast, simple, and reliable: the obvious interface of choice. As always with new technologies, it is wise to let some other poor fool rush in and do the beta testing: by sitting back a little from the bleeding edge of technology, you end up with very nearly equal performance, at significantly lower cost, and with vastly better reliability.

The 66MHz bus Pentium II chips were never very popular here (mostly a value for money thing — see the CPU guide), but this was one of the two boards we mostly used for them. These came in the already popular but inherently flawed ATX form, which we have never liked. That aside though, the P2L97 was a typical ASUS product: very fast, superbly documented, and a pleasure to work on.

  • CPU support: Pentium II 233, 2366, 300, 333, Celeron 266, 300.
  • Speed: 66MHz.
  • Slots: 5 PCI, 2 ISA, 1 AGP.
  • RAM: 3 168-pin SDRAM, up to 384MB.
  • Cache: None.
  • Chipset: Intel LX, Award BIOS.
  • Status: Legacy.
Elite P6LX-B illustration

Elite P6LX-B

Although you didn't hear their name much in Australia, Elite were a huge concern — one of the top five motherboard makers in Taiwan.

This was in many ways a very typical LX board: simple, solid and not too expensive. We preferred it to a host of otherwise similar boards because, as much as for any other reason, it came in the industry-standard baby AT form and thus could fit in an ordinary mini-tower case. For upgrades this was the only possibility, and even for new systems it was both cheaper and more reliable than ATX. (As a matter of fact, it still was, right up until AT boards became unavailable early in the new century—see our Shapes, sizes and trends section for more on this.)

Once the VIA Apollo Pro chipset for the Slot 1 processors arrived however, there was no longer much point in buying an LX-based board — a 100MHz capable Apollo Pro cost about the same.

  • CPU support: Pentium II 233, 266, 300, 333, Celeron 266, 300.
  • Speed: 66MHz.
  • Slots: 4 PCI, 2 ISA, 1 AGP
  • RAM: 3 168-pin SDRAM, up to 384MB.
  • Cache: none.
  • Chipset: Intel LX, Award BIOS.
  • Best With: Celeron-A
  • Status: Legacy.
FIC VA-503 illustration

FIC VA-503+

The first 100MHz Super 7 motherboard to hit the market was destined for an extraordinarily long life. These debuted as a rather expensive but very high-performance line, and gradually became cheaper, but the extra speed the 1MB on-board cache provided kept them at the performance leading edge for years.

Most 100MHz boards required 100MHz PC-100 SDRAM which, though difficult to find and quite expensive to begin with, soon became almost universal. The 503+ was an exception, because in amongst its ill-thought out and confusing collection of jumpers there was an option to let you step the RAM clock back to 66MHz but still run the rest of the board at full speed. The 503, in other words, offered three different clocks: the main clock controlled the board and cache RAM, the multiplier controlled the CPU, and the RAM clock was separate again. Naturally, you got the best performance with new 100MHz SDRAM, but if you were upgrading on a budget you could use up your old PC-66 or EDO RAM and still get most of the benefit of a new board and a K6-2. The speed of the main board itself, in other words, was more important than the speed of the RAM or, more correctly, that big fast 100MHz cache masked the slow main RAM so well that the speed loss was only a few percent. As the Pentium II had already demonstrated, RAM speed matters, but cache speed matters even more.

Not surprisingly given its amazing flexibility, the 503+ had an absurdly difficult jumper layout, but performance on the test bench more than made up for it. When these first appeared, CPUs like the Cyrix 6x86MX-200 and the Pentium MMX 233 were still common in new systems, and the VA-503+ was distinctly sluggish with them. But with the new generation chips, particularly the K6-2, they came into their own: with anything over 250 or 300MHz they were outstanding. It was the 503 that went into the vast majority of our higher performance systems in '98 and '99, initially the mighty K6-2/300, and then a long string of progressively faster variations on the theme: 350, 400, 450, and finally the doyen of Super 7 CPUs, the magnificent K6-III. Updated several times over its like, the 503+ and the K6 family provided the best higher-end price-performance solution on the market for nearly two years — an unprecedented run of success — and it continued to offer solid entry-level value for another two years after that.

  • CPU support: 6x86, 6x86MX, C6, K5, K6, K6-2, K6-3, P54C, P55C.
  • Speed: 66, 75, 83, 95, 100, 112 and 124MHz.
  • Slots: 3 PCI, 3 ISA, AGP
  • RAM: 4 72-pin FPM, EDO or BEDO and 2 168-pin SDRAM, up to 512MB.
  • Cache: Surface mount, 1MB pipeline burst.
  • Chipset: VIA Apollo MVP3, Award BIOS.
  • Best With: K6-2, K6-III
  • Status: Legacy.
QDI Advance 2 illustration

QDI Advance 2

These were our first QDI board for quite some time. The Advance marked a major change in QDI's policy: for many years they had been an Intel-only manufacturer. The modern PC market had become much more diverse, and manufacturers changed with it.

Entry-level or not, these still had an excellent feature set: the very fast VP3 chipset, AGP slot, good documentation.

While we have never been mad keen on jumperless main boards, given the industry shortage of entry-level boards at the time, we thought we'd give these a go for the low-end (mainly the IDT C6-225 and IBM 6x86MX-233). The Advance 2 used the VIA VP3 chipset — essentially the same as the leading-edge 100MHz MVP3 except that it was limited to 83MHz; not really a factor in the context of the non-100MHz capable CPUs we used these for. (To be more precise, the VP3 was the forunner of the MVP3.)

Alas, the Advance 2s were prone to having niggle problems, mostly CMOS and clock-related, and had a fairly high return rate. We switched to other boards before too long: mostly the MVP3-based ones from FIC and Epox.

  • CPU support: 6x86, 6x86MX, C6, K5, K6, P54C, P55C.
  • Speed: 50 to 83 MHz.
  • Slots: 3 PCI, 2 ISA, AGP
  • RAM: 4 72-pin FPM, EDO or BEDO and 2 168-pin SDRAM, up to 256MB.
  • Cache: Surface mount, 512k pipeline burst.
  • Chipset: VIA Apollo VP3, Award BIOS.
  • Best With: 6x86MX-300, C6, K6-266
  • Status: Legacy.