BOINC does an amazing job of farming out complex work units. Could we break these up into even smaller batches, so that users can realize the impact they're having on research?

I think this change would introduce a relatively minor overhead performance hit on the cluster, with the benefit of being more accessible to our userbase. It's more satisfying to see the work units processed counter tick up faster, even though the total processing rate is about the same.

Increasing the number of work units adds overhead to the server side of things. The whole idea is to deploy the work across the distributed grid of machines. Each work unit is already flexibly designed to allow as much effort as your machine can devote. It is how there is a user preference for work unit runtime. Longer runtimes mean more models (or "decoys") are calculated for each work unit.

I see you are running Linux machines. I don't believe the graphics work on Linux. But if you run on Windows or Mac, you can SEE the work unit being processed by opening the graphics or using BOINC as your screensaver.

You might also check out the fold it game they've created. It teaches you how the proteins work, and allows you to design your own binding molecules for the COVID-19 target.

---

Rosetta Moderator: Mod.Sense