Does anybody have access to any performance data for GROMACS on a Blue Gene/L system? I'm putting together a proposal for some time on such a machine, and some moderately accurate costings would be nice. Mark