To boot the epic, create a tst_bench_qtconcurrent with building blocks for later extension.

This first benchmark should allow to configure (or run a range of) work items of different (CPU, no I/O) duration: e.g.

• FP addition
• FP sqrt
• atomic int increase
• memcpy of different sized buffers
• ...
• QString::toLower() of differently-sized strings
• ...
• QImage::transform() (note this may use QThreadPool under the hood) of differently-sized images
• ...

And the total number of these items to run. This defines the working set for the benchmark runs.

The first actual benchmark should then compare the wall-clock execution time for running the working set in

• sequential
• parallel with multi-valued QFuture (resultAt) using 1..N threads

This is already a 3d data set, so one might want to visualize it in some clever way.