Parallel computation between Intel i7 and M2 pro

mesh
1

To deal with 3D optimisation problem in FEniCS, i use MPI to realise the parallel computation and try on two different computers:

  1. Intel I7 with 4 cores 8 threads which is 3GHz*8
  2. M2 Pro with 10 cores

the result is that a test process require 600s on M2 pro while it takes only 90s in Intel (i’ve make use of all cores for each computer).
Anyone has met the same problem?

2

How did you install FEniCS?
Are you using legacy FEniCS or FEniCSx?

3

Thank you for your quick answer.
Both two are legacy FEniCS version 2019.1.0,
the one in Mac is installed through a docker and
the other Intel one is installed directly by using Ubuntu packages.