Michael Seaton*; Luke Mason*; Zakhar A. Matveev†; Stephen Blair-Chappell‡ * STFC Daresbury Laboratory, UK† Intel, Russia‡ Intel, UK
This chapter introduces a tool that radically improves the ease at which you can analyze the nature of the vectorization in the hot loops of a program. The discoveries this tool helps make on how compiled code is vectorized, or what is stopping code from being vectorized, checking for loop dependencies, and observing the memory access patterns, are of value regardless of what system you are targeting. The examples provided in the chapter, clearly show what can be achieved using the Vector Advisor.
Lattice Boltzmann equation
Explicit vectorization ...
Get High Performance Parallelism Pearls Volume Two now with O’Reilly online learning.
O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.