RMK001: loop nesting that might benefit from hybrid parallelization using multithreading and SIMD

Issue

Use multi-threading and SIMD hybrid parallelization to maximize the parallel performance of the nested loop.

Actions

Parallelize the outer loop using multi-threading and the inner loop using vectorization. You can do so for OpenMP by invoking:

pwdirectives --multi omp-for+simd foo.c:5 -o foo-hybrid.c

Relevance

Certain nested loops may benefit from using a hybrid parallelization approach where the outer loop is parallelized using multi-threading while using vectorization for the inner loop. In this way, the use of modern hardware is maximized by using as many CPU cores as possible through multi-threading and at the same time using the vectorization hardware available in each core.