While at the recent Microprocessor Forum, we heard quite a few presentations about upcoming, soon-to-be-released processors. Because this event was the 10th Forum, the program also featured several retrospectives.

Putting the two types of talks together, we were struck once again by the lack of progress in the chips in X86-based servers--and by the opportunity this slow improvement rate has created for Unix vendors.

Flash back to the first quarter of 1996, and the hot server processors everyone was discussing were Digital's 333MHz Alpha, Sun's 167MHz SPARC and Intel's 200MHz Pentium Pro. Today, almost two years later, the hot topics are the 600MHz Alpha, which is nearly twice the speed of its earlier counterpart; the 300MHz SPARC, which is also almost twice the speed of the earlier chip; and the 200MHz Pentium Pro. Moore's Law sure hasn't done much for X86 server buyers during this time.

To be fair to Intel, we must point out that megahertz are not the only way--or in some cases even a good way--to measure performance. For example, Intel did boost the Level 2 cache on the Pentium Pro to 1MB, gaining a performance increase of about 10 percent to 20 percent. Still, that's not a lot of progress for almost two years.

Intel has also introduced the Pentium II (Klamath) in that time, and that chip is already available at speeds up to 300MHz. The Pentium II is not, however, a great choice for serious servers, because multiprocessing Pentium II-based systems can't easily contain more than two CPUs. Intel has scheduled relief for 1998--the Deschutes implementation of the Pentium II. That chip family will, like the Pentium Pro, support four-processor configurations and thus get X86-based processors back on the serious improvement track. Too bad Deschutes comes a full two years after the debut of the 200MHz Pentium Pro.

On the software and system side, the news for X86-based servers is both good and bad. The good news is that Windows NT is hot, and Microsoft is doing all it can to make NT even hotter. The bad news is that no matter what demos or scalability days Microsoft concocts, no one is yet convinced NT can seriously scale in systems with large numbers of processors.

During this same couple of years, several good things have happened for Unix servers. As we noted above, the processors in Unix servers have increased greatly in speed. Unix servers with four, eight and more processors have been readily available. The Web has skyrocketed, and the Unix-dominated Web infrastructure business has grown accordingly. All the pieces are in place for Unix server vendors to stage a huge assault on the market and really put Microsoft's NT plans and Intel's server strategies to the test.

Despite this opportunity, NT, not Unix, is the hot topic with almost every corporate buyer we meet. The Unix server vendors aren't getting the job done yet, but if they're smart they'll seize this opportunity to regain from Wintel a serious lead. If they don't, they will have no one but themselves to blame when their businesses begin to erode seriously, because this could well be their last opportunity.

Copyright (c) 1997 ZDNet. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of ZDNet is prohibited. ZDNet and the ZDNet logo are trademarks of Ziff-Davis Inc.