I am currently running MD simulations on a HPC.
Some simulations are running faster for a lower number of CPUs. In other words, sometimes I run the same simulation on 10 and 20 CPUs, and the simulation running on the 10 CPUs runs faster.
Can anyone explain how is that possible?
Thank you.
Some simulations are running faster for a lower number of CPUs. In other words, sometimes I run the same simulation on 10 and 20 CPUs, and the simulation running on the 10 CPUs runs faster.
Can anyone explain how is that possible?
Thank you.