[backend]

Parameterises the backend with

1. precision — number precision, note not all backends support double precision:

   single | double

2. memory-model — if to enable support for large working sets;

   should be normal unless a memory-model error is encountered:

   normal | large

3. rank-allocator — MPI rank allocator:

   linear | random

4. collect-wait-times — if to track MPI request wait times or not:

   True | False

5. collect-wait-times-len — size of the wait time history buffer:

   int

Example:

[backend]
precision = double
rank-allocator = linear