ZDNet heeft een artikel gepost over Anand's K6-2 review. Ze beschuldigen Anand ervan dat-ie pro-AMD is. Vreemd, want 2 week geleden was-nog een 'AMD basher'. Het is ook nooit goed.
Hier is het artikel:
Codenamed "Sharptooth," the chip is expected to be released in January. While most of the details of Sharptooth are public, AnandTech was the first site to publish benchmarks of the coming processor.
"It looks like Anand has an OEM source that is feeding him these chips," said Tom Kehoe, an AMD (NYSE:AMD) spokesman. Kehoe would not comment on the accuracy of the benchmarks shown on the page, but said the release caused some consternation at the Sunnyvale, Calif. company.
Anand Lal Shimpi -- the 16-year-old high-school student who runs AnandTech -- confirmed that the chip was sent to him by a PC maker. The review of the chip has netted the junior over 300,000 hits in the past 12 hours.
Topping the charts
If the benchmarks are accurate, AMD rival Intel Corp. (Nasdaq:INTC) could be in for a rough year. In the AnandTech review, an engineering sample of the 450MHz K6-3 topped the charts, outperforming Intel's 450MHz Pentium II and a Celeron A running at 450MHz.
How did AMD soup up Sharptooth? By adding a 256KB cache directly to the processor, similar to the tactic Intel used to turn its Celeron processor from an underdog to a top performer. In Intel's case, the company only added 128KB of secondary memory, but it still boosted performance by 20 to 30 percent in most applications.
According to AnandTech's tests, the K6-3 out performed a Pentium II operating at the same clock speed by as much as 12 percent, always beating out Intel's processors.
Chip analyst Dean McCarron of Mercury Research Inc. thought that the review could easily be legit. "Those numbers are not surprising, given some of the focuses that AMD has been working on," he said. However, McCarron was quick to point out that, even if the benchmarks are reasonable, AnandTech could easily have made up the numbers. "I don't think he did," said McCarron, "but that is always a possibility."
Still, the review lacks some key benchmarks. In fact, all the tests played to AMD's strengths in business applications or 3DNow optimized games. While AMD insists that the company had nothing to do with the reviews, the lack of standard Winbench 3-D and 3-D rendering tests, which frequently favor Intel's processors, makes for a glaring omission.
Anand said any bias was unintended, and that in fact, more strict CPU tests would be posted on his sight soon.
To be fair, since the chip tested is so-called "alpha" silicon -- that is, a pre-product -- the actual K6-3 chips that will come out in January could perform much better. "The difference between alpha silicon and final silicon is sometimes as great as 2x," said Mercury's McCarron.
Neither AMD nor McCarron thought that would be the case this time around, and Anand estimated that his benchmarks would be right on. "To the best of my knowledge, the primary difference between this sample and the actual product will only be stability," he said.
Still, it does look as though AMD has an early start on next year's competition.