Skip to content

[walker] use parallel_for instead of parallel_reduce

Tobias Leibner requested to merge walker_parallel_for into master

Currently, we are using parallel_reduce with an empty join function, which corresponds to a parallel_for if I am not mistaken. So this MR changes the walker to use a parallel_for directly. Any objections, @ag-ohlberger/dune-community? Probably does not matter much in terms of performance, though.

Merge request reports