Hello,
I have a question on running Turbomole on clusters with queuing system. Some modules are parallel and some are not. If user is running simulation with i.e. 48CPUs and TM is executing aoforce, my 47CPUs are wasted. Is there any way to make TM to push some other computations in that time? Or is it a good idea to split single TM script into parallel and serial parts and split them in PBS? Does it save something?
regards
Wojtek