Vectorization Essentials 03 - Vectorization II

Vectorization Essentials 03 - Vectorization II

Explores some concepts underlying vectorization. Explores the trade off between using compiler autovectorization versus explicit vector programming versus SIMD intrinsics and assembly. This video sets the context for using explicit vector programming. It compares explicit vector programming as being similar to explicit parallel programming using OpenMP parallelism constructs, where the developer takes control and responsibility for vectorizing specified loops. The video also gives quick examples of the three big ideas in explicit vector programming: SIMD loops, Intel Cilk(TM) Plus array notations, and SIMD-enabled functions

See other videos in this series:

Download Video

852X480 (22.02 MB)

For more complete information about compiler optimizations, see our Optimization Notice.