More openMP: find_min/max() and sum() #1449
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Originally in #1430, but as that needs more work, creating a separate PR. Sadly, in most cases (except LAFOV), the functions that I parallelised were pretty fast already.
There's more work to do
+=
etc, as it might be a bad idea to parallelise this forVectorWithOffset
, as then the nD-Array functions could create a lot of threads. I guess it needs specialisation forArray<1,elemT>
.Will likely be rebased on top of #1442 once that is merged, although I could merge as well as it's independent.
@markus-jehl I haven't parallelised
is_contiguous
, as that's also a bit more work, see e.g.https://stackoverflow.com/questions/9793791/parallel-openmp-loop-with-break-statement. Alternatively, we introduce an extra member (if it turns out to be worth it, as that'd occur a lot in current nD Arrays).