You can find the article at
LATimes.com, but you have to sign-up with the "free" registration.
For those that do not want to sign-up:
AMD Taking Gamble With 64-Bit Processor
Technology: Big businesses have shown a need for such powerful chips, but consumer demand may be weaker.
By MATTHEW FORDAHL, ASSOCIATED PRESS
For much of its 33-year history, Advanced Micro Devices Inc. prospered by making cheaper, sometimes faster versions of microprocessors pioneered by Intel Corp.
Now, AMD is planning its biggest launch ever--and it's no Intel knockoff. The new processors share some of the same elements found in powerful servers but will also target personal computers.
It's a big financial and technological gamble for the world's No. 2 microprocessor maker.
So far, AMD has failed to make much headway in the server market and has been struggling more than usual lately getting into personal computers because of stiff competition and weak demand.
If it succeeds with the new chips, developed not coincidentally under the code name "Hammer," AMD could become the microprocessor industry's powerhouse, setting the pace for computing for years to come.
But if AMD stumbles, Hammer will join a growing list of powerful chips nobody wants.
"We've established something that still needs to prove itself by getting into the market and succeeding," said Fred Weber, chief technology officer of AMD's Computation Products Group.
An Athlon-branded desktop version is scheduled for release early next year, several months late because of production delays. Chips for servers, branded Opteron, will follow later in 2003.
Hammer chips process data in 64-bit chunks rather than 32-bit chunks, as is the case with today's Intel Pentium or AMD Athlon chips. Currently, only high-end servers and workstations use 64-bit processors.
The increase in the number of bits is similar to widening a freeway, allowing a higher volume of data traffic inside the processor and to memory. Still, a computer's operating system and other programs must be optimized to take advantage of a 64-bit processor.
But unlike today's 64-bit chips, the Hammer chips speak the same language--the x86 instruction set--as today's 32-bit chips as well as yesterday's 16-bit and 8-bit chips.
"People want to be able to run their existing code," Weber said. "There are a lot of applications that don't need to move to 64 bits ever."
Intel's 64-bit offering, Itanium, abandoned the x86 instruction set in favor of Explicitly Parallel Instruction Computing. Intel says it was designed for high-end computing needs. Though Itanium can run 32-bit code, it loses performance.
AMD maintains that its Hammer technology will give users more options, particularly businesses that want to run 64-bit programs alongside existing 32-bit software without side effects.
But although big businesses have shown a need for such powerful chips with resource-intensive simulation and database programs, it's not so clear that ordinary consumers need them.
"If you look at the types of work people do, if you look at the data sizes in which they operate, 32 bits is where the desktop's at," said Louis Burns, general manager of Intel's Desktop Platform Group.
AMD concedes as much.
After all, it took a decade to switch from 16-bit to 32-bit computing. By the end, the PC world moved from text-only interfaces of MS-DOS to the graphics of Windows 95.
Initially, Hammer won't have a native, 64-bit Windows operating system. Until Microsoft Corp. develops one, as promised, users will have to run the 32-bit version.
AMD maintains that the Hammer chips will still have a performance advantage, even running 32-bit code, because they incorporate some functions typically performed by chips outside the processor. The result is a 10% to 15% improvement in speed over comparable 32-bit processors, said John Crank, AMD's senior branding manager.
AMD has yet to release key details, including clock speeds and their true performance compared with existing 32-bit chips. Nor has it released exact pricing, though it is expected to be in line with Intel's top chips.
"This is a classic AMD strategy of trying to find the transition point between product introductions for Intel ... and introduce products that fit right into those holes," said Eric Ross, an analyst at Investec Inc.
