Benchmarking 7.0 performance

[Geoff Canyon]

I'm wondering if anyone has benchmarked 7.0 performance? I did two quick
tests and found:

on mouseUp
   put the long seconds into T
   repeat with i = 1 to 20000000
      put "A" && "aa" into X
   end repeat
   put the long seconds - T
end mouseUp

Took almost 5x as long to run as on 6.7, which I expected since it's
presumably doing UTF things under the hood.

But I also found:

on mouseUp
   put the long seconds into T
   repeat with i = 1 to 20000000

   end repeat
   put the long seconds - T
end mouseUp

Took about 1.5x as long to run on 7.0 as on 6.7. Obviously our own code is
still a much larger factor -- "repeat for each" in 7.0 is still going to
handily beat "repeat with" in 6.7 for processing large chunks of text, etc.

gc