Systems-integration lessons learned
Systems-integration lessons learned
Andy Wilson Editor at Large
Editing a trade publication can be an interesting, rewarding, and sometimes surprising business. Last month, when Larry Brown, one of our contributing editors, told me of his intention to build a Pentium-based image-processing system, I was delighted. Not only would this put Larry on the leading edge, it would also teach him a few lessons about systems integration.
To accomplish this feat, Larry set out on a "how to build your own PC" course. After three weeks of instruction, the necessary knowledge was memory-mapped into Larry`s brain. Now it was time for the hardware. After purchasing a PA-2005 motherboard from FIC (Fremont, CA), the search for a CPU was on. Brown was told to call The Electronic Planet (Nashua, NH), a supplier of ready-to-install CPUs.
The CPU specified by the instructor was a 6X86P33GP from Cyrix (Richardson, TX). When plugged into the FIC motherboard, the processor did not soft-boot to the floppy drive. Being a skeptical hack, I immediately questioned the technical expertise of Brown and his teachers. I consulted Thomas Pabst`s Web site--http://sysdoc.pair.com, a venerable wealth of information on CPUs, motherboards, graphics, and more.
Choosing a motherboard
Under the Mainboard Guide, Pabst lays out the hard and fast rules of choosing a motherboard: check whether it supports Pentium-compatible CPUs such as the Cyrix/IBM 6X86 and M2 or the AMD K5 and K6 in its BIOS; check whether it supports the dual-plane voltage of the Pentium MMX, AMD K6, Cyrix/IBM 6x86L, and M2; and make sure it has a decent switching voltage regulator. That must have been it: the BIOS.
Before I came to this conclusion, I decided to check with Cyrix. A quick look at the company`s Web site (http://www.cyrix.com) indicated that the Cyrix Platform Verification Lab had itself tested this motherboard for 6X86 processor electrical and functional compatibility. And apparently, according to Cyrix, this includes detailed application-based testing using Winstone 96 and 32 benchmark suites under various operating systems!
To be fair to Cyrix, I decided to give them a call. Being a member of the press, I decided the fastest way to get results was to ask the public-relations department. Surely, they would route me to the correct technical member of staff at Cyrix. I e-mailed Cyrix`s public- relations department, which was very helpful. They told me they would ask the relevant people and get right back to me. That was more than three weeks ago.
Meanwhile, back in the lab, Larry was still checking and setting jumpers. But the CPU still didn`t recognize the A: drive.
Rather than do the honorable thing and take back the Cyrix part and refund the purchase price, staff at The Electronic Planet claimed they could not because it was used. They would only refund 80% of the money.
So, Larry drove to PC Build (Needham, MA). After replacing the Cyrix part with a 90-MHz Intel Pentium, the motherboard worked perfectly. One of the technicians there described the Cyrix part as "quirky" and said he did not recommend it anymore.
Finally, after much frustration, the Pentium motherboard, which is to be the basis of the image processor, is operational. All that is needed now is a low-cost PCI-based frame grabber, a digital CCD camera, and some good image-processing software. I wonder if anyone has ever built any such system using a PA-2005 motherboard. It may help Larry build his system this year as opposed to next.