That's perfectly fine and you can up it to 999 seconds which is about 17 minutes. The effect is though that any checkpoint save writing is delayed to the next one occurring after the write delay interval, but, what is 5 or 10 minutes of computing time loss when a system is stopped/restarted?

Added: And no it has no effect on memory use. Any checkpoint occurring during the delay is simply skipped. The science works on the basis of: Can i write a checkpoint *now* when its scheduled to do one and the client talking back with yes or no. The 60 seconds is default for BOINC. Some projects write a checkpoint very frequent and others few. See Start Here forum for checkpoint saving for more information.

Less disk i/o is better performance. Particular if checkpoint saves are large will it effect overall wallclock time for a work unit and *could* it be noticeable on lower end machines. The key is oodles of RAM else on large workunits would there be substantial Ram to swapfile i/o. Just look at what the maximum ram use is for a project and set the work and idle to a level that it does not impair regular use. The more ram allocated to BOINC sciences, the better.

The BOINC screensaver, if it works, has an option to go blank after x minutes. This allows to show off the picture to the interested for a few minutes and after goes off for maximum idle while crunching while idle.