OPNG_0119947_00096
Yes, crazy "job" numbers in these WU's, and pathetic credit/points.

OPNG_0120000_00007

OPNG_0120128_00091

OPNG_0120843_00118

@bluestang (and others)


Unfortunately, the project folks probably have no idea when this sort of thing is going to happen -- the jobs are batched up into tasks using an algorithm that attempts to estimate the amount of effort needed per job based on the nature of the ligand and the number of dockings requested! If it is still using the original algorithm the tasks it is building appear to be built appropriately - however... :-(

It was reasonably good for OPN1, although it doesn't allow for different receptors and grid sizes; at least with the CPU tasks as the time goes up so does the credit... However, it doesn't make anywhere near as good a job of building GPU tasks, as we can all see (and, of course, the application is using a different docking algorithm - there's a bit about it in this news item .)

I have been monitoring the ligands and receptors for which I've been receiving work in the hope that I might work out some adjustments to suggest to get a more consistent work-load; bigger grids often lead to longer run times and there are other oddities too - however, every time I think I've spotted something useful, a batch of ligands will show up to contradict what I though I'd found :-(

I don't think this is likely to improve at the moment (the person responsible for the original work hasn't been seen here for ages now; presumably he's one of the folks who has already left the IBM team...) and it remains to be seen whether there will be any will to do anything about it once the transition to Krembil is completed. Who knows, perhaps someone will actually manage to work out a better "scoring" algorithm for individual jobs that can be easily plugged in to replace the current one... We can dream, can't we? :-)

Cheers - Al.

P.S. whilst some of us don't really care about the credits, it would be nice if the tasks had a more realistic set of jobs (usually quite a few less with the current target!) as, apart from anything else, it would mean more tasks for those who struggle to get regular work! And, given the way the credit is calculated, reducing the number of jobs in a task would not automatically reduce the credit earned...