9.
Next time you do this, you should exclude the benchmark results that are a combination of sub-results. For example, Webmark has two sub-benchmarks, commercial transactions and information processing. The overall Webmark result is simply the geometric mean of these two, it is not an independent measure of performance, so it should be excluded from the overall geometric mean or Webmark will be double weighted. The same thing is true of PCMark04 and its two sub-benches and maybe some others.
IMO, you should also totally can any synthetic benchmarks from the overall geometric mean. Why should az user care if he can read or write his memory faster than his neighbor? So, not only should the overall PCMark04 be eliminated, but all its CPU and Memory components also.
Also, why not include a Celeron M chip and the new Dothan chip. People want to know how well these chips perform at something other than Mobilemark and Business Winstone.
[
Posted by: Petz

|
Date: 05/19/04 09:39:59 PM]