Moore's Law: True, But Irrelevant?
Page 1 of 1
Moore's Law, which states that the power of microprocessors will double roughly every 18 months, has been a constant in the technology industry since 1965 when Intel co-founder Gordon Moore first coined it. But is Moore's Law still relevant when information technology (IT) shops consider when -- and whether -- to replace mainstream desktop PCs?
That's the core question in a debate that has broken out within the computing industry and the analysts that observe it. Virtually everybody agrees enterprise servers and PCs used for specialized, processor-intensive applications always will require ever-more power. Similarly, there's widespread agreement that chipmakers can continue abiding by Moore's Law into the foreseeable future.
However, in the past, increasingly powerful software created a need for more powerful PCs, which, in turn, led to software that was more powerful still. Many believe that cycle -- and the IT purchasing decisions based on it -- are no longer relevant.
"There is very little in terms of mainstream software that is coming out anytime soon that needs more performance on the desktop," said Mark Margevicius, research director at Gartner, Inc.
"We recommend that corporate buyers look at Office 2003 because it has better collaboration capabilities that add to productivity," he said. "And that will require new desktop capabilities."
In a time of tight IT budgets, this debate has strong ramifications for enterprise buying decisions. That, in turn, means the stakes are enormous for PC manufacturers, chipmakers and large software vendors.
IT Managers Weigh In
IT managers we spoke with tended to agree with Margevicius.
"We definitely have some power-hungry users, but the chipmakers haven't sold me on the notion that I need more powerful computers for most of my users," said Greg Gomez, IT director for McKee Wallwork Henderson Advertising in Albuquerque, N.M.
Outside of his company's graphics department, which uses Macintoshes, Gomez said he has no plans to purchase new desktop PCs users for at least another year. Most of his company's existing desktop PCs already are between two and four years old, he said.
Gomez said he isn't interested in upgrading to Microsoft Office 11, which he called "just more of the same. My users use the same features they've been using since Office 97 -- they're comfortable with that feature set." The company's current crop of desktops work perfectly well with Office 97, he said.
"The vast majority of my hardware purchases this year are going toward servers," Gomez said.
Angelo Tomasello agreed. He's director of information solutions for Interactive Business Systems, an Oakbrook, Ill.-based IT consulting firm.
"My clients realize that Moore's Law may be true, but they don't care," he said.
Many PC upgrades in the past were motivated as much by laziness as by need, Tomasello said.
"Once on-site staff runs into complications, particularly operating system instability, they often will say 'This system has been in place for three years, so we should replace it,'" he said. "Often, they don't bother diagnosing and fixing the problem."
Now or Later?
The bottom line issue for IT shops is how often they should replace mainstream desktop computers. Margevicius believes that the clear answer is: Not as frequently as once was the case.
"Sometime around the Pentium 3 there was a major switchover," he said. "Until then, software had always outpaced hardware and you had to buy mainstream PCs more often just to keep up. Now, the hardware outpaces software." In addition, he said that PCs are more reliable now than they once were.
As a result, Margevicius, "We recommend that clients can now safely go four years without replacing desktop machines." Previously, the recommendation was to replace desktop PCs every two-and-a-half to three years.
Gomez concurred, while Aberdeen Group's Kastner disagreed.
"It used to be we bought new desktops every time we upgraded our applications," Gomez said. "That's not true at all these days."
"Windows 98 and NT machines are obsolescent," Kastner asserted. "It's pretty provable that workers with those machines can't make use of the productivity improvements in the latest software."
Kastner cautioned IT managers against trying to save money by lengthening the time between desktop PC upgrades.
"Over the long haul, that's penny wide and pound foolish," Kastner said. "I don't say that every person in an enterprise needs a $3,000 desktop, but more do need one than many people say."
Everybody agreed that Intel's new hyperthreading architecture, which is now becoming available for servers, will indeed be attractive for mainstream PCs. That technology enables simultaneous processing on multiple threads.
"I turn my PC on in the morning and there's 40 processes going on, even if I'm only checking my e-mail," Kastner said. "That's eventually going to be important for desktop PCs."
In the meantime, however, each enterprise must determine how relevant Moore's Law is in terms of its desktop PC buying decisions.