In this PR, the allocation of memory in DiscreteMorseSandwich has been parallelized using OpenMP tasks.
This results in a speed-up of x3.65 on 24 cores (the computation step goes from 2.45s to 0.67s on average).
The results were verified using Pierre Guillou's pdiags_bench benchmark.
The same type of parallelization was performed on the Discrete Gradient's memory allocation step.
Hi all,
In this PR, the allocation of memory in DiscreteMorseSandwich has been parallelized using OpenMP tasks.
This results in a speed-up of x3.65 on 24 cores (the computation step goes from 2.45s to 0.67s on average). The results were verified using Pierre Guillou's pdiags_bench benchmark.
The same type of parallelization was performed on the Discrete Gradient's memory allocation step.
Thanks for any feedback.