1999: no change

Gigabyte GA-5AA

A nice Super 7 board. As usual with the ALI Aladdin chipsets, these were a little slower than the benchmark-setting VIA MVP3-based boards, but still in the ballpark.

Like most Gigabyte boards, these were well laid out and beautifully presented. Some extra care had gone into the finishing touches: putting a couple of cents worth of black plastic around the serial port headers, for example, made life much easier when you were working in a crowded case and it is something we wish all board manufacturers would do. Notice that although the board was a little light-on for slots, they were all usable. (Compare with the Epox EP-58 below, which had two extra slots but restricted their length with the CPU socket and which led to cable snarls because the PCI and AGP cards contended with the serial and printer port outlets.) The new-look AMI BIOS was very nice — AMI had finally gone back to the excellent user interface they favoured in 386 days, so it was as easy to use as Award at last.

Setting the 5AA up was easy but (a consistent Gigabyte failing, this) the over-copious documentation managed to carefully avoid telling you what the switches actually do. If you're trying to set one of these for a CPU that is not specifically mentioned in the manual, you have to work everything out from first principles. When these were new this was no problem, as all the common CPUs were listed, but it was easy to predict that in six or twelve months time it would become an issue. In summary, a worthy board but not fast enough to push FIC and Epox off our list of favourites.

In retrospect, the 5AA lasted remarkably well. It remained an excellent product right up to its dying day, and our respect for it increased with the discovery that the 5AA supported all the odd-ball final variations on the K6 family that appeared in the second half of 2000 — notably the K6-III+ and the K6-2/550. The only other board we know of to do this is the VA-503 (with an undocumented beta BIOS flash for the K6-III+) but with the 5AA you just plug in, set the jumpers and switch on. No BIOS update was required.

In later years, however, the 5AA dated quickly. Where VA-503+ based systems remained practical and trouble-free for years, the 5AA's ALI chipset became more and more problematic (largely because of IDE driver issues and a flawed AGP implementation). You normally expect the workload of a mainboard to increase over time (as people switch to more demanding applications and add things like CD burners and faster video cards) but the 5AA did not cope with aging gracefully.

• CPU support: All current Super 7.
• Speed: 66, 75, 83, 95 and 100MHz. Unofficial 105, 110, 115 and 120MHz support.
• Slots: 3 PCI, 2 ISA, AGP
• RAM: 3 168-pin SDRAM, up to 768MB. (256MB cacheable.)
• Cache: Surface mount, 512k pipeline burst.
• Chipset: ALI Aladdin 5, AMI BIOS.
• Best With: K6-2
• Status: Legacy.

Epox EP-58 MVP3C-M

Yet another VIA MVP3-based 100MHz Super 7 board. We had never tried Epox boards before this one, and were impressed by it. Presentation and layout was excellent, right up to Iwill standards, and like recent Iwill boards, the EP-58 used the best possible jumper arrangement: one for bus speed, one for CPU multiplier, and a third for voltage. (There is a larger view of the follow-on Epox MVP3C that shows this more clearly.)

It was a shame that they put the jumper block at the front right corner of the board, though, because that part of the board is often very difficult to see or reach, particularly in desktop cases. Still, there was plenty of room for a full-length PCI card without fouling the CPU fan. It was a bit of a toy feature really as you couldn't access it without rebooting in those days, but if you went into the advanced CMOS setup there was a real time CPU temperature readout. Everything does this these days, of course, but that was the first one we saw and we thought it very cute.

If the EP-58 had had had a full 1MB cache like the 503+, it would have been almost perfect.

• CPU support: 6x86, 6x86MX, C6, K5, K6, K6-2, P54C, P55C.
• Speed: 66, 75, 83 and 100MHz.
• Slots: 4 PCI, 3 ISA, AGP
• RAM: 3 168-pin SDRAM, up to 384MB.
• Cache: Surface mount, 512k pipeline burst.
• Chipset: VIA Apollo MVP3, Award BIOS.
• Best With: K6-2, MII
• Status: Legacy.

FIC VA-503A

Broadly similar to the evergreen 503+, these sacrificed a couple of ISA slots and EDO RAM sockets, but added chipset-embedded VIA sound and an audio modem riser socket. We pondered introducing these for entry-level systems and eventually decided against it. So, it seems did a lot of other dealers, as they never seemed to make it to Australia!

The chipset was the VIA MVP4, essentially an MVP3 with sound added. Later on, we were to get to know the MVP4 better and like it not much at all: see the Microstar M-5187 entry, and compare with the SiS-based ASUS P5S-B. When it came to integrated chipsets, SiS were still the best.

The provision of only a single ISA slot was a potential problem. Intel and Microsoft were hell-bent on eliminating ISA altogether: we didn't think this was wise in 1999; perhaps five years later. Out here in the real world, there were countless millions of people with a need for ISA cards, and ISA had more than sufficient bandwidth for all but the most speed-hungry devices. We have industrial customers with practical, working ISA cards that would cost $2000 and $3000 each to replace with a PCI device, and in some cases, there is simply no alternative to an ISA product. Or, in the home and small business world, take a look at the price of a second printer port card, for example: you could get them in PCI but they were horribly unreliable and cost three times as much. Even as late as 2002 they were still fifty percent dearer, and to this day they usually don't work properly.

Notice the small brown connector at upper right, between the PCI and ISA slots. This was one of the first occurrences of the Audio Modem Riser slot, the AMR — one of the most asinine inventions ever to take up valuable mainboard space at the expense of useful things like PCI and ISA slots. In the three or four years since the AMR and its various bastard children first appeared, we have never seen one in actual use on a real computer. (We saw one on a Compaq once, with a modem in it, which in itself is evidence for the prosecution. Very little that Compaq do in their mass market machines is notable for anything except low cost of manufacture. Needless to add that the Compaq had come in to have its modem problems sorted. Once we threw away the AMR device and fitted a Rockwell external, it worked perfectly.)

• CPU support: M-II, K6-2, K6-3.
• Speed: 66, 75, 83, 95, 100, 112 and 124MHz.
• Slots: 4 PCI, 1 ISA, AGP, AMR
• RAM: 3 168-pin SDRAM, up to 768MB.
• Cache: Surface mount, 1MB pipeline burst.
• Chipset: VIA Apollo MVP4, Award BIOS.
• Best With: N/A
• Status: Legacy.

FIC KA-6100

This was the first motherboard to use the VIA Apollo Pro 100MHz Slot 1 chipset and was incredibly long-lived. In the early days, it offered near BX performance at less than LX cost. We sold quite a few of these in 1999 and came to trust them: they were cheap, stable and very flexible.

But only a mother could have ever loved them: they came with a three year warranty, a voluminous and detailed manual, all the proper drivers on a well-organised CD-ROM — and a host of individually trivial but collectively infuriating detail design mistakes. None of them were serious, but we wished FIC would think a bit harder about the poor bloody technician.

The documentation was comprehensive but badly organised, with lots of clear but useless diagrams and detail about the obvious, but tiny, cramped little diagrams of the things you actually need to know, hidden away in the middle somewhere where you couldn't easily find them.

The board design was like that too: it was needlessly difficult to locate pin 1 on the cable headers, or work out which of the IDE connectors was the primary; the cables for the (optional) on-board audio card were too short, and the pin-out for the auxiliary cables (speaker, power LED and so on) was simply bizarre. To make matters worse, the printed diagram was upside down! (A well-laid out board doesn't need a diagram, the pattern of the pins tells its own story.) The funny thing is that FIC used to do superb board layouts a few years before this: products like the PA-2005 and the PIO-3 were simple and obvious at a glance.

Notice, however, that none of our complaints about the KA-6100 relate to the actual function of the board: they ran perfectly (though hardly at the speed of light even by 1999 standards), they were just a pain to work on. And we liked their speed, their price and their low return rate: they were an excellent board once you bolted the lid on and handed the system over to your customer. Just don't ever expect one to be easy to install!

• CPU support:Celeron and Pentium II/III, all speeds.
• Speed: 66, 75, 83 and 100 MHz.
• Slots: 3 PCI, 2 ISA, 1 AGP
• RAM: 3 168-pin SDRAM, up to 384MB.
• Cache: none.
• Chipset: VIA Apollo Pro, Award BIOS.
• Status: Legacy.

FIC VB-601

With a host of BX mainboards to choose from, why pick this particular one?

There didn't seem to be a great deal to choose between BX boards after they had been available for a while—usually a good sign that the market is mature — and the VB-601 was as good a choice as any. FIC's traditional reasonable pricing and their three year warranty did it no harm, and while the documentation was rather patchy, at least the setup was straightforward. We didn't get to know these very well though, as they sold in tiny numbers; during their time almost everyone was buying K6-2 or Celeron. For the cost difference to upgrade to a BX board instead of a KA-6100 you could go another CPU speed grade instead or buy more RAM, and either way get a noticeably better system.

• CPU support: Celeron, Pentium II/III.
• Speed: 66 to 100MHz.
• Slots: 5 PCI, 2 ISA, AGP
• RAM: 4 168-pin SDRAM, up to 1GB.
• Cache: N/A.
• Chipset: Intel BX, Award BIOS.
• Best With: Pentium II/III
• Status: Legacy.

Shuttle HOT-591P

Another of the bumper crop of 100MHz Super 7 boards that 1999 produced. Like the VA-503+, it was based on the outstanding VIA Apollo MVP3 chipset.

We had been very impressed with Shuttle boards of this period and this was no exception. The layout was excellent, with simple, clear jumper settings; much better than the arcane VA-503, though still not up to Iwill standards. Unfortunately, Shuttle followed the all too common practice of sticking all the minor switch and LED connectors in a single block with poor labelling. (That's it at lower right. Most older boards had each connector separate so that it was easy to work out which was the reset, which was the PC speaker, which the power LED, and so on.)

The CPU socket position meant that you couldn't use a long PCI card like a Voodoo II. The glossy manual was tiny but had all the necessary information, and some really nice touches, like the neat set of stick-on labels for printer, modem, and so on. Performance was very good, though not quite as fast as the class-leading VA-503+.

• CPU support: 6x86, 6x86MX, C6, K5, K6, K6-2, K6-3, P54C, P55C.
• Speed: 66, 75, 83 and 100MHz.
• 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, 512k pipeline burst.
• Chipset: VIA Apollo MVP3, Award BIOS.
• Best With: K6-266 and 300, K6-2
• Status: Legacy.

PC Chips M571

Veteran Red Hill readers will be astonished to see us featuring a main board from these one-time champions of sleaze. Frankly, so were we! This was the manufacturer that gave the world several of the worst motherboards ever made, including the notorious 486 boards with fake cache RAM and the dreadful VX Pro Pentium boards.

Back around Christmas 1998 there was a major shortage of entry-level Socket 7 boards and we ended up with a batch of unbranded SiS-based boards with integrated audio labelled (rather misleadingly) "PC-100" — and being a 66MHz main board, PC-100 they were not! Having nothing else available, we tried them out with some Cyrix 6x86MX-233 CPUs and they seemed to go rather well, so we bought some more. By the time we discovered who they came from, we'd sold 60 or 80 of them.

In practice, they were very similar to the old ASUS SP-97V or the Elite P5SJ-B, but with an odd little Soundpro audio chip. The strappings were clear and logical, though you had to remember to open the "clear CMOS" jumper which, for some reason was set closed when they shipped. Rather to our surprise (we never quite trusted anything that wasn't a real Sound Blaster) the Soundpro chip seemed to go well too.

At the time we wrote: “So are still using the PC-100 in entry-level systems, and we have yet to return a single board. They are cheap, easy to work on, fast enough for this sort of work, very reliable, and great value for money. We can't quite believe we are happily selling a PC Chips motherboard, but how can we argue with a zero failure rate?”

• CPU support: MII, C6, K6-2 (with 66MHz bus).
• Speed: 50 to 75MHz.
• Slots: 4 PCI, 3 ISA
• RAM: 2 168-pin SDRAM, 4 72-pin fast page or EDO, up to 384MB.
• Cache: Surface mount, 512k pipeline burst.
• Chipset: SiS 5597, AMI BIOS.
• Best With: 6x86MX-233, MII-300
• Date: 30th March 1999.

Update: June 1999. A new revision of these boards arrived on the market late in their life, and they were as dreadful as any of the PC Chips boards of old. They looked almost identical, but came with cute multi-coloured I/O cables and a "Y2K Ready" sticker on the BIOS. We had gone through eleven boxes of ten M571s up until then, and not had a fault with even one of them. Of our twelfth shipment, the new revision, eight out of the ten gave trouble. The old M571 was bulletproof, but you'd be wise to avoid these like poison.

Oh — and don't worry, this was a one-off. We know when to quit, and we have not the slightest intention of ever buying any other PC Chips products.

PC Partner VIB878DS

Not one of our regular boards, but quite interesting. PC Partner are one of those firms that seem to have an ordinary reputation around the traps, though we've never been quite sure why. The occasional PC Partner boards we've seen over the years have been perfectly functional, and though we've seen the odd one fail, the failure rate seems to be about the same as the industry average. Sure, there are brands we like better, but we can certainly think of brands we like less too. The only time we have used any great number of PC Partner boards was back in Pentium days, when we had a batch of 30 or 40 PC Partner VX boards. They were typical VX stuff: nice and simple, pretty reliable, well-priced.

The 878 came to our attention because of a shipping mistake — we got a box of five when we'd ordered a different brand. So, of course, we had to try them out.

The idea was reasonable enough: a standard MVP3 chipset with a Sound Blaster 16 chip added onto the board. The implementation was extraordinary: the board is tiny — not much bigger than a typical 386SX, and so crowded that it just had to come from Hong Kong! This alone made it difficult to work on, though an intelligent layout saw to it that you could still fit your full quota of PCI and ISA cards — even full-length ones. The jumper layout, however, was one of the worst we have ever seen — possibly even more awkward and illogical than the FIC KA-6100 above. As regular Red Hill readers will know, we don't like jumperless main boards — the less you trust to software, the more reliable your system is — but products like this really made us wonder.

They were reasonably priced and they seemed to go OK, but they would have to have been a lot cheaper than they were to make it worth the extra trouble.

• CPU support: 6x86, 6x86MX, C6, K5, K6, K6-2, P54C, P55C.
• Speed: 66, 75, 83 and 100MHz.
• Slots: 4 PCI, 2 ISA, AGP
• RAM: 2 168-pin SDRAM, 2 72-pin fast page or EDO, up to 256MB.
• Cache: Surface mount, 512k pipeline burst.
• Chipset: VIA Apollo MVP3, Award BIOS, Sound Blaster 16.
• Best With: tweezers and a lot of patience. Not great at 100MHz.
• Status: Legacy.

Epox EP-V370B

A very nice Socket 370 board, laid out with typical Epox care, and beautifully documented. We switched to these as our standard Celeron board towards the end of 1999 and had no cause to regret it. As we pointed out above, the old Apollo Pro chipset was a little slower than a BX, but the cost difference was more than enough to take you to the next CPU speed grade up — so it was well worth it. (BX zealots won't be satisfied with this, of course. But then BX zealots are never satisfied.)

These were an example of that rare type of board that techies like most of all: you take them out of the box, you plug them in, and they work, so you forget about them. In our book, that's pretty close to perfect.

They wouldn't win any speed prizes, but they were fast enough for most uses, and although we didn't sell all that many Celerons (K6-2 and K6-III were both more popular here), we never needed to send even a single EP-370B back for warranty replacement.

Just one gripe: when would the manufacturers finally start shipping USB cable connectors with motherboards as standard instead of as an optional extra? It saved them maybe one dollar and it was way past time.

• CPU support: Celeron, Socketed Pentium-III (if they ever arrive).
• Speed: 66 to 100MHz.
• Slots: 3 PCI, 2 ISA, AGP
• RAM: 3 168-pin SDRAM, up to 768MB.
• Cache: N/A.
• Chipset: VIA Apollo Pro, Award BIOS.
• Best With: Celeron
• Status: Legacy.

FIC SD-11

In the flesh, these were huge — the biggest motherboard we had seen since early 386 days. The SD-11 was FIC's first Slot A board for the AMD Athlon CPU, and together with two others, the only board that was suitable for the Athlon when the chip was first released. As a high-end product, these didn't sell in large numbers to begin with, but the Athlon's performance was awesome.

Of the original three Athlon boards, this was the only one to use the VIA south bridge chip: the other two (from Gigabyte and Microstar) both used the AMD one — not that it seemed to make much difference in practice. There seemed little to differentiate between the FIC, Gigabyte and Microstar Athlon boards: they were all ATX and any performance difference was tiny. Over time, as the VIA chipsets took over, AMD gradually phased out production of the 750 chipset in order to concentrate on CPUs.

Early on, the cost was quite high as motherboards go, but dropped gradually as the Athlon moved into the mainstream, and eventually stabilised at about the same as a BX board, or perhaps a little less.

One thing we didn't like about the SD-11 was that it only had a single serial port — there was no point in this oddity (at least not that we could see) and it was enough to encourage us to try out a couple of other brands where we might otherwise have stuck with the one we'd come to trust. A second feature — or design blunder, to use the correct term — was the odd-ball positioning of the board-edge connectors for printer ports and USB and so on. For some incomprehensible reason, FIC choose to put them in the wrong order so that you can't use any of the several common ATX case-plates. They shipped a special one with each SD-11, which you can just pop in instead of whichever one is fitted to your case, but if you loose it you have trouble — the only resort is the traditional tin opener and sticky tape fix. Not a clever bit of design, FIC.

• CPU support: Athlon.
• Speed: 200MHz.
• Slots: 5 PCI, 1 ISA, AGP
• RAM: 3 168-pin SDRAM, up to 768MB.
• Cache: N/A.
• Chipset: AMD 751/VIA VIA 686, AMI BIOS.
• Best With: Athlon (what else?)
• Status: Legacy.