To add, such assignments would be subject to availability at time of work-call i.e. a host may not always get the most optimized work for that device.

--//--

Hello WilhelmGGW
Reference: WilhelmGGW [May 24, 2012 9:18:40 PM] post

If you are interested mainly at machine performance or efficiency
----------------------------------------------------------------------------------
I think that it is technically possible for the WCG-servers to provide the infrastructure-support to do the clientLoad-to-clientDevice matching to come up with an optimized project WU distribution. At this point, that sounds costly though.

If you won't mind the extra effort entailed, for now, you can have all the data you need to make decisions as to what among your computers crunches what WCG project for a criteria (most-efficient, most performance, etc). As soon as you have the data, you can manipulate your project-settings, profiles, etc to match the project offerings from WCG. This calls for a lot of work from your end -- and much of that is because there is no automated way to do it, for now, at the client-side.

If a project's "value" is more important
-------------------------------------------------
Some people don't care much about the machine's performance or efficiency. As long as it's not too bad, these people are fine with a project that they see as having value, significance, or meaning. Because these people also contribute to the grid effort, part of the incentive to focus on purely machine performance or efficiency is thereby diluted. Combine this with the server-side cost in supporting clientLoad-to-clientDevice matching, and I'm afraid that you will have to do with the manual method... for now.
;