PERF: Improve convolution performances
Improve convolution performances
Summary
Improve convolution performances by 5%-7% on big filters
Rationale
Convolution can be quite inefficient in computation time.
This refactoring:
- helps the compiler factorising out a test done in a computation loop
- replaces a division by a multiplication
The speed-up measured is between 5% and 7% on thousands elements wise filters.
Additional notes
Along the way an exception specification has been removed. ITK has already removed these deprecated annotations, a 10ish remain in OTB and should be taken care of eventually before trying to compile OTB with C++17 compilers.
Copyright
The copyright owner is CNES and has signed the ORFEO ToolBox Contributor License Agreement.
Check before merging:
- All discussions are resolved
- At least 2
👍 votes from core developers, no👎 vote. - The feature branch is (reasonably) up-to-date with the base branch
- Dashboard is green
- Copyright owner has signed the ORFEO ToolBox Contributor License Agreement