I would have a look in your event log for clues as to why you're not getting as much work as you'd like.

Seems to me this was set towards the end of HCMD2 because of the totally random lengths and run-times. Many systems were overcommitting with very small cache sizes.

As for the limit, I only rarely hit it and that is when running only FAAH Vina.

Doesn't seem to be to burdensome.

Right.

Unless you want to load a bunch of tasks and abort all the VINA till you have enough Autodock which should carry you through.

There's a conflict between how boinc is designed to work using the duration correction factor and wcg having disabled this function remotely for the version 7 agents. The dcf for these is locked to 1.000000 for the project, speak wcg overall. Some projects do this to prevent buffer overload because of higher variable runtimes from workunit to workunit and batch to batch. Devices that process faster than average rarely hit buffer setting and devices that are slower often have more work than wanted. Regrettably, this and adjusting the buffer to that behavior also affects the work request frequency, when using maximum additional work buffer and the order of processing, the earlier deadline first, if set over a certain factor relating to deadlines. All in all, loading multiple days and only connect once or twice a week is problametic in amongst. Everyone has to work out for themselves what's the best setting to suite their needs, far from optimal.