That's odd. I get a different file from following your link now, than I did earlier. I think the file I get now is the one you where complaining about. Your complaint made no sense at all in terms of what was at those pages numbers when I first followed the link.
(Edit - Maybe its possible I made a mistake and went to the wrong page number, but I'm pretty sure I typed in the correct number)
Looking at this other file - Its not just one benchmark where AMD has a lead. And they do show Intel as having the overall lead in some of their charts. See figures 4, 5, 10 for Intel having an overall lead. See figures 6, 8, 11, 12, 13, 21, 27 for other benchmarks where at least the top AMD chip has a lead.
As for using a geometric average there isn't anything wrong about that, so the only real claim is that one specific benchmark (the one with the biggest AMD lead) is bogus.
And it may indeed be bogus, but you don't make much of an argument to back up your contention.
Figure 22 is a display of the benchmark for a benchmark called "High Performance Gaming and Multimedia Performance". Looking at the "Benchmark Methodology" section of the PDF, it seems that that benchmark is a combination of running "City of Villains" and Windows Media Encoder 9. I suppose the reason why you claim that network latency is the issue is because City of Villains is being run online, but while that makes it possible that network latency is primarily what is being tested, it doesn't make it certain. It might be worthwhile to try the same chips on different mother boards and different NICs (if they use NICs rather than build in ethernet) and see if you get different results.
The benchmark is obviously chosen to boost AMD's scores, but whether or not its entirely bogus can't be determined from the information in the PDF or from any information or argument you have presented. |
