Real Advantage

Specifications

The Problem

Let me prefix this project by stating, I love my AST computer, if you go read my article about it you'll understand why, but TL;DR, It was the first Pentium machine I ever owned.

However, even as a teenager, I always felt that the machine was... lacking. And when browsing the manual and other material for potential upgrades I found myself running into roadblocks at every turn.

During my first ownership of the original machine, I upgraded from a Pentium 75 to a Pentium 100 and added some more RAM for a total of 16MB.

My major gripes were the buggy chipset and/or ASTs implementation of them, their "My first X" style BIOS, and I understand it was aimed more at first-time buyers but they could've at least added an "advanced" mode or menu to give people like me more control and lastly the lack of PCI slots which limited you quite a bit on expansion.

But what else was available in the same year? Surely not all motherboards were created equal? Indeed they were not.

The best laid plans

For this project I had acquired an AOpen AX5T motherboard since it was from around the same time that the AST machine was produced.

Designed as a bridge between the aging Pentium 90 and the older 486s, the AOpen AX5T-3 represents what was possible in motherboard design of the era. The board supported everything from the P100 all the way up to the much newer 233 MHz Pentium MMX, 200MHz Pentium MMX OverDrive, and AMD K6 PR2-233. It also included the bleeding edge, USB 1.0, which was ratified in the same year, showing just how on the ball AOpen was with its design.

However, I had tested the board several weeks ago and it was functioning perfectly, so I left it on my open-frame bench case, but when I went to test it again last week, it was refusing to POST, the complaint appeared to be "No video card detected" according to the beep codes.

After trying dozens and dozens of configs, CMOS clears, BIOS flashes, I gave up and ordered a POST diagnostics card, since I didn't have one to hand.

When it turned up, I plugged it in to the PCI slot and flipped the power on.

For a moment, nothing.

Then I could see and smell smoke, I turned it off post-haste but it was too late, the damage was already done, the 12V rail feeding the PCI slots on the back of the board was completely destroyed. The trace runs the entire width of the board and runs through multiple vias and many other components. It would not be a simple "bodge wire" fix. It just simply wasn't worth repairing because even if I did. it still had the POST error which, in my opinion, would most likely be the BGA chipset joints as the board was ever so slightly warped from being stored in a box all those years, and fixing it down to the test bench probably but a lot of stress on it resulting in a broken solder joint somewhere, and I don't have the tools to reflow BGA parts, yet.

So I took to the Retro Web and searched up a similar board. While not a perfect match, the Soyo SY-5EMA was close enough and even though it's a super seven it still supports Socket 5 CPUs down to 100MHz, just like the AX5T.

Better, Faster, Stronger

For the remainder of this, I will be comparing the Soyo as if it had been released in 1996 just like the AOpen and AST boards. So while the Soyo does come equipped with a an AGP slot which in and of itself a huge plus for massively expanding the available graphics adapters that can be used with this board as fast PCI 3D accelerators (that weren't 3DFX) are hard to come by. I won't be using the slot while the board is in use for this project to ensure a fair fight as the AGP standard wasn't ratified until around 1998.

With five PCI and two ISA slots, the Soyo easily accommodates most expansion cards from a previous 486 or earlier Pentium machine, depending on how heavy of an ISA user you were or how much you relied on integrated devices. This stands in contrast to the AST Advantage's two PCI and three ISA slots, which limited future expansion options quite a bit since newer cards were likely to be PCI.

AST's also had other issues, Its rather slow SiS chipset and limited CPU support - only up to P166, with no support for the OverDrive 200 or MMX processors - really constrained its useful lifespan. And while the chipset claims to support chips above 166, I have never been able to get of them working with any available version of the BIOS. An implementation problem on ASTs side? SiS also supported DMA transfers but the feature was essentially unusable - and enabling it in the drivers prevents Windows from booting, at least on this motherboard, so again could it just be incorrect use of the chipset by AST? In contrast, the SY-5EMA's VIA MVP3 chipset offered robust Ultra-DMA 33 support from its launch due to thorough QA and proper design methodologies, resulting in reliable, high-speed data transfers that made a real difference in day-to-day use.

Perhaps if AST hadn't been going out of business, there might've been a BIOS update to address these issues. Unfortunately, we'll never know.

Expansion

While AST had everything "on-board", The SY-5EMA takes a different approach to integration. There's no VGA nor sound, which will cost you two slots - however it grants substantial freedom to choose your own hardware without worrying about resource allocations and conflicts or having to disable on-board equivalents to use add-in cards.

Another smart design choice was the inclusion of both RAM slot types. This gave users the ability to use their existing EDO/FPM RAM from previous machines to start with, then upgrade to SDRAM later when money allowed - definitely an economical upgrade path for money-savvy users.

The AX5T-3's inclusion of USB headers shows its forward-looking design. Released in 1996, the same year USB 1.0 was introduced, it was ready for what would become the dominant peripheral connection standard. The AST Advantage, released in 1995, predated USB's introduction entirely - a stark reminder of how quickly PC technology was evolving during this period, but they would've known what Intel was about to propose, and instead of delaying or allowing room for a USB card to be added, they shipped it anyway. Adding USB to the AST would've consumed 1 of 2 precious PCI slots.

By 1998, the SY-5EMA already had USB headers and ports available on the board as standard, indicating just how quickly USB had become the dominant peripheral connection method.

Planned spec

For this build, I'm thinking of using a Matrox Millennium 2 for graphics, a Sound Blaster Live! (CT4760) for audio, 64 MB of PC133 SDRAM and a 100MBit NIC for data transfers.

As for fixed storage, I'm going to try and acquire another Quantum Atlas V 9.1GB just so that the AST and this machine are on equal footing, while the drive is from 1999, finding SCSI drives from 1996 is proving to be a difficult task.

I do have a Western Digital Caviar 1.6GB IDE drive which is period correct, and is what the AST would have originally been equipped with, but I feel 1.6GB would be too limiting for my experiments.

I have to admit though, I do absolutely love the sound the Caviar drive makes.

Thoughts so far

The Soyo SY-5EMA motherboard will prove to be a reliable and versatile choice for building a machine of the same era as the AST while providing much more flexibility and choice. It's a shame that my AX5T board can't be used for this project, but the Soyo is a solid, comparable choice. Its VIA MVP3 chipset, while not an Intel chipset, with its stable drivers, and hardware compatibility makes it an attractive alternative. With a configuration in mind, I'm excited to see how the SY-5EMA performs in actual experiments. Meanwhile, I'll continue to search for a 1996-era SCSI drive, hopefully one that fits the project's requirements. I'll update this page again in the future as I progress, please check back soon!

Back to top