Hello Ian,

We are seeing what we think are severe memory leaks with our BBH simulations using the current trunk. Memory usage goes up linearly during the run, then drops to its original value after recovery. Does anyone else see this?

I don't know about trunkg but as far as I know Jim Healy had things like that. I also some time ago had issues with that and found that HDF5 was caching (HDF5) metadata. There should be code in trunk by now that disables caching though. So my first try would be to disable all HDF5 output (maybe excluding checkpoints if you only write a single one).

Yours, Roland