Not that I expect that the folks at Macworld magazine are technical rocket scientists, but this latest piece of "lab" work has me puzzled as to what they're trying to prove and who they're working for.
In a series of "tests" performed on the iMac Core Duo and the iMac G5, the "labs" tested the following:
- iMovie 6.0 effects rendering and export
- iPhoto import
- iPhoto export
- Create ZIP archive
- iTunes Rip
- iDVD burn
- iSquint conversion
- A battery if BBEdit commands
Now, they were using all new Universal binaries, so they theorized that they should all be 2-3x as fast for all the operations, just like Steve said. Well, of course they aren't, but the reason they aren't is much more likely linked to the type of operations that are being done than the processor. Because, as you can see by looking at the list for everything except the last item in the list, they are all disk I/O intensive operations, and the disks haven't gotten any faster or slower in the new machines.
Hopefully, they'll think about doing some processor benchmarks one of these days...
[Follow-up] In the long line of comments on this post at MacWorld there are a few references to the non-CPU portion of the benchmark, including a pointer to a comment on one of Apple's external (user) mailing lists stating that XCode compilation on the iMac Core Duo is almost as fast as the same on a PowerMac G5 Quad.