December 27, 1996 1:15 PM PST
PCs: The next generation
One of the most talked-about developments in 1996 was MMX. This technology for boosting multimedia processing will finally make its way into the market via two Intel chips: the P55C Pentium, due to be announced January 8, and the next-generation P6 family of processors known as Klamath, due in the second quarter.
"The P55C would have been a great kicker for fall sales...Intel began producing the P55C processor in the late fall but chose not to announce it until January to minimize the impact of the Christmas season," said Michael Slater, publisher of the Microprocessor Report, in his upcoming December 30 edition.
"Klamath initially will boost P6-family pricing to levels Intel enjoyed in [the first half of 1996], but the company is likely to bring Klamath prices down rapidly as it moves it into the mainstream. By early 1998, Deschutes (the code name of a processor to follow Klamath)--with clock rates of 300 MHz and higher--will take over the high-end performance price points," said Slater.
Slater expects the 233-MHz version of Klamath to be priced at close to $700, while the 266-MHz version should come in at just above $800. This puts both of these chips in high-end, above-$3,000 systems when they are introduced in 1997.
"Existing Pentium Pro systems won't be easily converted into Klamath and will continue to sell for some time. A 200-MHz Pentium Pro may also outperform a 233-MHz Klamath on applications," said Slater. The reason for this odd set of circumstances is that the current Pentium Pro processor has a very-high-speed, built-in cache (memory chip), but in the upcoming Klamath this will not be built in order to reduce cost.
"It seems likely that Intel will position Klamath as a desktop and low-end server [processor] initially, with Pentium Pro focused on midrange and high-end servers. Sources indicate that Intel is not supporting Klamath for systems with more than two processors, so multiprocessor systems will remain a Pentium Pro stronghold until Deschutes ships," Slater said in the report.
But all this processor activity doesn't necessarily bode well for notebook PC users looking for a next-generation processor. These users may have to wait for the next next-generation Deschutes processor in order to get a wholly new notebook PC processor.
"The P55C will be Intel's fastest mobile offering in 1997; Klamath's power consumption will be too high for reasonable notebook offerings. Although the P55C offers a modest boost [in performance] from the existing Pentiums, it leaves a significant gap between high-end notebook and desktop performance. Deschutes will run at a lower voltage and enable a mobile P6 processor. Deschutes won't be available in production until late 1997, however, and it won't be until sometime in 1998 that a mobile Deschutes processor will be available," said Slater.
The Pentium processor, of course, will also be affected by a rapidly changing market. "The 100-MHz Pentium will all but disappear by midyear, while the 133-MHz [processor] takes the economy spot and the 166-MHz version becomes the mainstream entry-level chip," he writes.
According to Slater, the availability of upcoming Intel processors should pan out as follows:
|Klamath||P6 with MMX, external L2 cache||Q2 1997|
|Deschutes||Smaller version of Klamath||Q4 1997|
|Katmai||Enhanced with MMX2||first half of 1998|
|Williamette||Next-generation x86||second half of 1998|
|Merced||First Intel 64-bit processor||first half of 1999|
"Debate is intensifying around network computers...This debate is not just another overblown battle of industry egos. Where, when, and at what rate NCs do or don't take off in the market--and how well PC and software suppliers anticipate and position their offerings--will largely dictate who the next leaders will be in microprocessors, end-user devices including PCs, and software," says International Data Corporation analyst Frank Gens a in recent report.
"It is interesting to see how the rhetoric of the 'PC vs. NC' combatants has shifted in the past year. Larry Ellison of Oracle championed the cause for low-cost, easy-to-manage non-PC user devices designed to leverage the Internet and private intranets--trashing the Microsoft-Intel PC architecture throughout. At that same forum, Microsoft's (MSFT) Bill Gates scoffed at the utility of any device offering less than full Windows functionality," Gens says.
"While each camp maintained much of their previous position, there was a perceptible shift, with each acknowledging the other side's view. There appeared to be some level of convergence on a vision in which both PCs and NCs will both exist and, to some extent, compete. And each side admitted that neither the PC nor the NC visions were, in themselves, complete.
Industry bigwigs Gates and Compaq CEO and president Eckhard Pfeiffer "outlined the need to create a 'manageable PC' with features that enable centralized management and lower costs of ownership. This concept was recently formalized with the announcement of the NetPC specification.
"The NetPC initiative is proof that the PC world is already anticipating the challenge from NCs. We expect some PC manufacturers--notably Compaq and HP--to reshape the traditional PC into a broader range of offerings with different form factors, functionality packages, and price points.
"We expect Intel [to] announce strong support for the NC concept, despite its strong stake in the PC market. Intel's decision to actively evangelize for NCs in addition to PCs will be a watershed event for the development of the NC market.
"Why will Intel support the NC heretics? Among PC technology suppliers, it is clear that Intel has the most to gain (and the least to lose) by endorsing the creation and proliferation of many new types of end-user devices. Intel is in an excellent position to leverage its strong position in the PC market into a leadership role in a segment that within ten years could dwarf the PC business. If Intel does otherwise and abdicates the NC business to other microprocessor suppliers, it will be at risk of being marginalized in the coming decade.
"Today's black-and-white 'PC vs. NC' debate will yield to a reality that includes a broad spectrum of information access devices ranging from set-top boxes and screen phones through very high-performance workstation PCs."
Gens also notes that one of the "most surprising concessions came from Scott McNealy, responding to a Pfeiffer question: "'When will users see a base of Java (i.e., NC-friendly) applications comparable to the large base of PC-compatible applications?'" Amazingly, McNealy took off his Java evangelist hat, and said it would probably take application developers "three years to burn through the $100 million Java fund and another two years to get to market." In this case, five years translates to 2001.
|related news stories|