News Focus
News Focus
Followers 4
Posts 4127
Boards Moderated 0
Alias Born 03/06/2003

Re: dougSF30 post# 28268

Sunday, 03/07/2004 1:22:00 PM

Sunday, March 07, 2004 1:22:00 PM

Post# of 98358
Keith, Doug, AthlonXP 3000+ speed grade -

Although I agree with Keith on the importance of effective product differentiation, I disagree that this can be used as a reason for the speed grade labeling. I'll have to be a stickler here and go back to the definition: A suite of benchmarks published by AMD which compare against the performance of a 1GHz T-Bird. The benchmark results meet or exceeds a linear extrapolation from the performance rating.

(Even when AthlonXP started falling behind Northwood, AMD did not violate this definition. They did seem to cut the margin, or underrating, enjoyed by earlier processors. Also, it can be argued that AthlonXP ratings did not always match Northwood clock speed: This is meaningless because AMD's ratings did meet their benchmark-based definition.)

Thus both an AthlonXP 3000+ and an Athlon64 3000+ must meet or exceed 3x the performance of a 1GHz T-Bird, as measured by the published benchmarks. AMD cannot maintain credibility if they start using the '+' rating differently for AthlonXP vs. Athlon64 without first offering an alternative measurement which can be independently verified.

I am not arguing that a 64-bit chip should be measured by the same standards as a 32-bit chip. Rather, I am arguing that AMD has set the measure, and so far they are measuring both using the same benchmarks. Thus they have to be held accountable for their own definition.

(I do think that there should be a set of 64-bit benchmarks to measure Athlon64, maybe AMD will do that when Windows AMD64 is productized. Perhaps they are waiting until they are satisfied that 64-bit compilers have reached maturity, but I understand they are doing pretty well now, on Linux at least.)

So if an Athlon64 3000+ outperforms an AthlonXP 3000+ on the published benchmarks then the only reason which is credible is that the Athlon64 3000+ is underrated.

P.S.: Kap, I see you said it much more succinctly -

AMD always said that model numbers represent relative performance of AMD parts _not_ a comparison with P4 or Celeron.


Volume:
Day Range:
Bid:
Ask:
Last Trade Time:
Total Trades:
  • 1D
  • 1M
  • 3M
  • 6M
  • 1Y
  • 5Y
Recent AMD News