I am sure that this has been covered somewhere but searching the fora I haven't been able to find it. I have seen reference to 30,000 computers running CPDN. At the same time the server status page currently has over a quarter of a million tasks in progress. That would suggest a lot of multi-processor macines! Is this figure a meaningless artifact? Is it just so high because of the number of tasks sent out that the system hasn't given up on yet?

Idle curiosity I know but the answer would be interesting.

Dave

Is this figure a meaningless artifact? Is it just so high because of the number of tasks sent out that the system hasn't given up on yet?

I think that's it. There are perennial problems with CPDN deadlines - and this is one of them. Set the deadlines too long and it takes a long time before the results are marked as expired; set them more realistically and multi-project crunchers complain that CPDN tasks are pre-empting other projects.

The fundamental difficulty is that climate models are big. If projects were very long and if the problem weren't topical then the length of the models might not matter. But a PhD is three years; grant allocations probably the same or shorter; IPCC AR4 was 2007, AR5 2013/14. It's hard to balance personnel, bureacracy, criticality, size and add some other factor of your own ...