By bigger I hope you mean longer, I did the too big whinge last weekend ;p

Perhaps an option to crunch long or short tasks could be introduced, with long tasks counting as two or three tasks, depending on their size (might inherently serve as an equilibrating method for different task types, and re-substantiate the existence of the task complete stats). The servers could automatically work out which task lengths to send to which computers, based on performance/turnover. This might facilitate improved server side distribution; it is easier on the server to package and send lots of tasks as larger tasks at the one time than lots of small tasks many times every day. However, if the option was available (automatically or by selection) to crunch smaller tasks it would also facilitate existing crunchers with more limited systems and extend their stay; a facilitation bridge until their next system.

While next year will bring new high end CPUs such as improved Extreme i7's and Bulldozer, we will start to see more mainstream mulitcore/thread systems (sandy-bridge cores will become more widespread). also
Ditto for the Zambezi, "In the middle of the second quarter, two models will have a total of 8 cores. The former has a maximum TDP of 95 watts and is therefore likely to be clocked lower."

While these may offer improved designs, many of the CPUs arrive with similar or lower clock numbers. So we are moving towards more cores but by in large not towards more speed per core (for the average system). Maybe something for the techs to think about in the new year, before these more-core CPU's come out. While multithreaded tasks may be less productive, many tasks could be packaged together/combined and split on the host.

I share anhai's concern about WU lengths specifically for C4CW. I teach in a university and have been introducing students to volunteer computing, specifically WCG. Many interested students have only low-end laptops or just netbooks, but they want to get involved and do what they can.

Currently C4CW and CEP2 are the only environmental projects on WCG. Obviously CEP2 is out of the reach of low-power laptops. So for kids who are particularly enthusiastic about environmental science, C4CW is their only WCG option. Target 02s already run 11+ hours on an Atom-based netbook; doubling that would be a bit discouraging for kids to whom a day seems like a long time.

Getting young people involved in WCG is such a win-win. It's educational for them; they get to participate in important science; their crunching does contribute immediately; and if they get hooked on crunching now, they're likely to contribute more later when they have more resources to put into computing.