Probably no point in even posting this....
The benchmarks you have referred to are generated using After Effects and Cinebench. After Effects does video processing, and Cinebench is a synthetic test that approximates performance. After Effects is one application of many used in video processing. Video processing is not all graphics programs.
As far as processing speed I could show what numbers a G5 generates running RC5 vs. a P4, check the listing of keyrates here:
http://www.orange.co.jp/~masaki/rc572/ratee.php
I could use that to conclude that for number crunching I'd be better off using a single G5 machine than 2 or 3 P4 machines. I could also use Seti@home to conclude the opposite. But it's only one application....
Likewise I could show how much faster the P4 was than the Athlon 64 doing MP3 encoding using Lame and conclude that for audio encoding P4s were faster, or the opposite by only looking at Ogg Vorbis encoding.
This is why when you see good gaming machine reviews, they benchmark the systems using 5 to 10 different games. If the above reviewer wanted to prove anything about Graphics Apps in general, I should have seen Photoshop, Illustrator, Lightwave, Maya, Media 100, Avid Studio, and quite a bit more. Likewise had they wanted to they could have picked a starting product and an ending product and done it using different programs on differing systems but achieving the exact same result (which is generally not done).
All that a benchmark proves is that a given machine can run that application at a given speed. This is why I always discounted Apple's famous Photoshop bake-offs and why I never paid much attention to any benchmark other than gaming. The reason for this is that machines are so damned fast these days that I rarely spend much time waiting on anything that's not dependent on network speed. Most of the time taken on the machine is taken up using the interface in order to get the machine to do something. If it takes 20 steps on one system, and 5 on another.....
Games are the exception for me for the most part. My gaming experience is actually affected fairly strenuously by the machine I'm using. If I'm playing America's Army, the higher the resolution I run the game at the easier it is to snipe the enemy. Also if the game gets really choppy during a firefight, I cannot fire at, out-maneuver, or turn to face my enemy before they take me down. This affects my experience in playing the game.
For other things.... I wrote a shell script that created over 18,000 folders, assigned permissions and ownership, and made them out in a specified hierarchy giving every student and teacher on campus a set of folders to manage assignments and other files. It took about 2 hours to generate the script and 5 minutes to run on a dual 1Ghz G4. Faster machines would not have helped, except for the SASI (student info database) server which is a dual Xeon with loads of RAM. I created a 100+ Meg Photoshop file with about 800 layers and took it to Kinkos for printing. Attempted to open on the "fastest" machine in the store (a 2Ghz P4). After waiting for 15 minutes I force quit Photoshop and walked over to a Mac (1Ghz) and had it open in 5 minutes to the amazement of the staff. This is so subjective as to be stupid. It does not prove anything other than perhaps that Kinko's Windows machines are all screwed up. PDFs are viewable on my old 733Mhz G4 using Preview as my 1733Mhz Athlon using Acrobat. I'd probably use preview on the Athlon except it doesn't exist. I play Dawn of War on Windows because there's no Mac port. I also play Doom 3 on the Athlon because the Mac version requires a much faster machine than the G4.
I'm interested in CPUs and discussions of such, however when it degrades into a battle of male insecurity with little regard to constructive discussion, I begin to lose interest.
