This is like a soup opera for tech geeks, keep it up you guys...

Well, it's not suprising that specialized processors, such as those gpus, put out a lot of work. Though, IIRC, the group behind folding@home has indicated that the more specialized processors aren't capable of doing certain tasks, so every group (including the inefficient cpus) will always have their part to play.

Do keep in mind, though, that the total number of units completed compared to total number of users doesn't mean all that much. The GPU users have a much lower ratio of active to inactive users, and that may indicate they are a much more dedicated group. It is possible that they complete more units because that demographic is more devoted.

@Squilliam IBM has processors other than the cell, it's true, I have read it in books.

There will always be a market for serious HPC people who don't want to run everything on mass amounts of crap commodity hardware tied together with a wet string. BlueGene/Q is targeted for 10 PFLOPS which seems to be significantly more than all of those machines currently running folding at home put together.

Also, I should note that Roadrunner is currently considered the worlds fastest computer features a hybrid architecture with 1/3 opterons and 2/3 Squilliam's arch-nemesis the cell.