Many OS/2 users who have been unable to decide between a Pentium- or Pentium Pro-based computer have had their decisions made even more difficult. Intel recently continued its march of market domination with the release of its new "Pentium II" CPU. Known to many by its preproduction code name, "Klamath", this newest speed demon from the world's largest desktop CPU maker promises once again to stave off the wolves (IBM, Cyrix and AMD) nipping at Intel's market share heels. It does this primarily, of course, by being fast, fast, fast. But it also gives Intel some breathing room by virtue of its new single-edge contact (SEC) cartridge design which will force many motherboard manufacturers to rework their products and will probably cause Intel's competitors to scramble to modify their own CPUs.

But it's not the fancy new hologram-bespangled packaging of the Pentium II that should concern OS/2 users, it's what's inside -- or rather, what went into what's inside.

To understand why I have reservations about a chip that will most likely run OS/2 faster than anything else on the planet, we have to consider a bit of recent history:

Throughout its existence, Intel has introduced increasingly faster generations of its CPUs at fairly regular intervals. By and large, these new generations were introduced with exorbitant prices and targeted at money-laden corporations that just had to get their numbers crunched 25% faster. As competitors came close to matching Intel's best chips' performance (or sometimes bettering it), Intel would slash the price of its flagship product and introduce the "next great thing". As the former high-end chip dropped in price, it became widely adopted as the industry standard.

The above story ignores some market factors, but it describes most of Intel's CPU product history. With one significant exception.

Late in 1994, the competition was gaining ground on Intel's Pentium CPUs so Intel was readying their new class of product, the Pentium Pro. This chip was a great step forward in many ways, borrowing design characteristics from RISC technology, bettering the speed of previous processors, and including a ton of L2 cache right on the chip. And, most significantly for those who had already joined the '90s, due to its new architecture, it was especially adept at running 32-bit operating systems and code. In short, if you were using a real 32-bit OS, this chip was the fastest thing going.

At the same time that Intel was finishing the Pentium Pro's design though, Microsoft was "finishing" its newly renamed operating system, Windows 95. Microsoft had long been promising that Win95 would be a great leap forward in both interface and architecture, notably calling it a 32-bit OS. Naturally, Intel -- Microsoft's most ardent bed-partner -- was set to capitalize on the release of this wonderful new OS with the release of their new chip which would run 32-bit OSs very well.

Except, of course, the hype from Microsoft was -- ahem -- a little less than completely accurate. Windows 95 was largely a step forward and certainly contained much new 32-bit code, but early on in its life, reports came flooding in that it contained chunks of old, pre-Win95 (read: 16-bit) code. This was proven conclusively when Win95 was run on Intel's powerhouse Pentium Pro chips. The new 32-bit chips choked on the new "32-bit" code.

Now, choked is a relative term, of course, and by no means did Win95 run slowly on a Pentium Pro -- nothing runs slowly on a Pentium Pro. But Win95 ran more slowly on the Pentium Pro than on the plain old Pentium. As you can imagine, this was a big problem for Intel.

Intel might have just held its breath and waited for the world to move to Windows NT (which did run faster on the Pentium Pro than on a classic Pentium); Microsoft certainly told them and everyone else that we all were moving there. Instead though, Intel saw the writing on the wall. The market was moving (like it or not) to Win95 and not to WinNT, at least not in any hurry. If Intel wanted to remain at the top of the heap, they had to come up with a chip that would run 16-bit code even faster than its current Pentiums. MMX technology is an offshoot of this need; the Pentium II is another.

And as we all know, most users of Intel-based PCs are still running either DOS/Win3.x or Windows 95. They are not running a wholly 32-bit OS and, therefore, have never purchased Pentium Pros in droves. For the first time in Intel's history, its flagship CPU has failed to become the CPU of choice for the mainstream market. In fact, the Pentium Pro has been available since November 1995 and has now been supplanted by the Pentium II but the majority of computer users still have never seen one in action.

Since this is the computer industry, Intel can't sit still and wait for consumer OSs to catch up with its chips. Two years is a lifetime, as they say, and some other chips such as the AMD K6, have matured quickly, and are ready to challenge Intel's top of the line CPUs. Intel realized some time ago that they had to make a move quickly, and they did.

Hence the Pentium II. In many ways another great leap forward from Intel. But in some ways a leap backwards too. Intel needed to improve the speed with which its high end chips ran Win95, and since Win95 is not wholly 32-bit that meant temporarily forgetting about all the 32-bit advances of the Pentium Pro and going back to the 16-bit drawing board.

Intel did (of course) succeed in retrofitting the Pentium Pro architecture to handle Win95's legacy code in a more speedy manner. According to PC Mag (vol. 16, no. 11), the new 266-MHz Pentium II performs roughly half again as fast as a Pentium Pro when running Win95 tests (designed to measure 16-bit code execution in the CPU) and faster than either a classic Pentium or a Pentium with MMX.

But what does Intel's research and development on the Pentium II provide for OS/2 users? Nothing. First, Intel invested time and money improving its chips. Then, they invested more time and money moving them backwards, retrofitting them to run 16 bit code faster. This isn't going to help many OS/2 -- or WinNT -- users.

The fact is, Microsoft's failure to live up to its claims that Win95 would be fully 32-bit have hindered Intel's ability to move the computing industry on to a better and faster chip design. Instead of moving forward wholeheartedly with a great chip (the Pentium Pro), Intel was forced to divert some of their energies that could have been better used to further improve the 32-bit code execution of its products.

And it's not only Intel that is focussing its energies on making chips to run an antique operating system. By all reports, AMD's K6 chip is comparable to -- if not faster than -- a Pentium II when running Win95. In other words, it is squarely aimed at knocking off Intel's Win95-centric products, not the Pentium Pro line. The entire industry is obsessed with running a 16-bit OS faster and is less concerned with running 32-bit OSs faster. This is the state of the art in the late 1990's?

Again, don't misunderstand me. I realize that the new Pentium II roars with speed on Win95, WinNT or OS/2 -- 16-bit or 32-bit OSs -- and I realize that it is good for us as well as good for Windows 95 users. It's just not as good as it might have been if Intel had not been forced to refocus its energies.

This is just one more sad example of how Microsoft's market dominance has prevented the computer industry from accomplishing what it could have.

--

Copyright © 1997 - Falcon Networking