There appear to be no benchmarks for QtConcurrent and QPromise and the existing QFuture benchmark doesn't really test the overhead of QFuture over, say, sequential execution or std::packaged_task, or, for map() etc, C++17's parallel STL algorithms.

Write a benchmark that allows to estimate the per-element overhead of QFuture/Interface/QtConcurrent::map/filter, compared to a) sequential execution and b) parallelized execution using b1) parallel STL and b2) std::promise/std::packaged_task.

If we find we're much slower, create follow-up ticket for optimization opportunities.