Advertisement
Published in:
2021 | OriginalPaper | Chapter
9. Vectorisation of the Model Problem
Author : Tobias Weinzierl
Published in: Principles of Parallel Scientific Computing
Publisher: Springer International Publishing
Abstract
With an understanding how computers handle floating-point calculations, we vectorise our model problem. For this, we first categorise different parallelisation programming techniques, before we commit to a pragma-based approach, i.e. rely on auto-vectorisation plus vectorisation hints to the compiler. Compilers can give us feedback where they succeed to vectorise, which helps us when we introduce OpenMP pragmas to our code. We wrap up the vectorisation topic with a discussion of showstoppers to vectorisation, such as aliasing, reductions, or function calls.
