A step-by-step guide to parallelising code

The article includes the before and after code, so you can clearly see the changes he made. He also talks through some of the optimisations he attempted that didn’t work, giving a valuable insight into the process of optimising performance.

He used OpenMP, and there’s an OpenMP getting started guide here, and he made use of Single Instruction Multiple Data (SIMD) instructions to speed up some of the core computations. There’s a quick introduction to SIMD here.

Whatever your code does, you might pick up a trick or two by reading this candid case study that reveals what worked, what failed and why.
This is part of the Modern Code section of Intel Developer Zone, providing resources to help you to optimise your software for optimal performance.

  • Show Comments