After a week or so of positive reports on Windows 7 pre-beta performance - with testers generally finding the upcoming OS, even in its current early state, besting Vista in real-world challenges such as start-up and friendliness to ageing or low-power hardware - we're perhaps overdue a more critical one. InfoWorld's Randall Kennedy certainly delivers the goods, with a full-on benchmarking comparison that, on the face of it, suggests that Windows 7 "threatens to deliver zero measurable performance benefits while introducing new and potentially crippling compatibility issues."
Of course, make a statement like that and people will start examining your methodology. Some of Randall's key principles - the "nearly identical thread count" and comparisons of memory use - are being picked apart amid accusations that they're relatively meaningless. His benchmarking, too, has prompted criticism, with commenters pointing out that video encoding is more a test of CPU prowess, while others question the validity of benchmarking non-final code at all.
Personally, while I'm sure Randall has good reason for his conclusions, I'm finding myself more swayed by the spate of netbook and UMPC installs we've seen that show improved performance over Vista. Perhaps that's a less scientific way of judging things, but certainly more relevant to the final critics themselves: end-users who get Windows 7 on a newly-purchased PC.