Thread,
Haven't seen this yet (but sorry if already posted). T'would seem to contradict the model-hurtz scheme idea (which is IMHO a good thing -- model-hurtz being a bad thing).
JJ
http://home-investing.excite.com/news/zd/010829/11/amd-charges-intel
AMD charges Intel with 'devaluing' the megahertz
Updated 11:21 AM ET August 29, 2001
by Ken Popovich, eWEEK
Intel Corp. is "devaluing the meaning of megahertz" with its Pentium 4 design, charges rival chip maker Advanced Micro Devices Inc., which claims the chip's design makes Intel's frequency claims misleading.
Just over a year ago, AMD proudly proclaimed its success in beating Intel to the 1GHz level with its Athlon chip. But now, following Intel's release of a 2GHz Pentium 4, the smaller chip maker is crying foul.
AMD representatives this week sought out reporters attending the Intel Developers Forum in San Jose, Calif., to raise awareness of the issue.
In particular, AMD contends the design of Intel's Pentium 4 has made the long-established practice of comparing CPU frequencies irrelevant.
"Over the last 20 years, AMD and Intel had similar architectures … so the fundamental difference between AMD's and Intel's products were the frequencies," said Aaron Seen, a spokesman for AMD, based in Sunnyvale, Calif. "Recently, with the advent of the Athlon and Pentium 4 processors, we've fundamentally diverged in terms of our architectures."
Specifically, Seen alleges, AMD's Athlon does more work per clock cycle than the Pentium 4. As such, an Athlon could essentially handle processing applications faster than a similarly clocked Pentium 4.
For example, the AMD representative said, the chip maker's 1.4GHz Athlon beats a 1.7GHz Pentium 4 in a number of benchmark tests, such as Business Winstone 2001.
In place of merely comparing clock rates, AMD proposes uses a technical formula to rate processor performance, essentially multiplying the number of instructions per clock times the frequency.
"We need to help educate the end users about the difference between frequency and performance," Seen said.
But AMD faces an uphill battle to change people's perception, one analyst said, and its focus on jargon such as "instructions per clock cycle" won't appeal to users outside the industry.
"Trying to do a comparison based on instructions per clock cycle, that's a tough one to get across to non-technical people," said Nathan Brookwood, principal analyst with Insight 64 in Saratoga, Calif.
Misleading frequencies
However, Brookwood agrees that Intel's Pentium 4 frequencies are somewhat misleading.
"It's a legitimate issue," he said. "The Pentium 4 is as much as 20 percent less efficient clock per clock than the Pentium III."
Essentially, he said, what AMD needs to do is make its point somewhat the way manufacturers of sports cars do.
"They need to focus on delivered performance: How fast does the car go with this engine in it?" Brookwood said. "If they can demonstrate that the car with the AMD engine in it can go just as fast or faster than the car with the Intel engine, you don't need to focus on what the RPMs were or the clock cycle speed is."
While AMD Athlon and Intel Pentium III were locked in a neck-and-neck megahertz race through much of 1999 and 2000, Intel has held onto the speed crown since the introduction of a 1.5GHz Pentium 4 in November.
Currently, AMD's fastest Athlon tops out at 1.4GHz, but the chip maker reportedly has plans to release a 1.7GHz by the end of the year. But by that time, Intel is slated to release a 2.2GHz-plus Pentium 4 chip. |
