Chapter 6

GPU programming

Abstract

Since GPU computation burst into the scene in the mid 2000s, it has taken the high-performance computing by storm. GPUs offer a leap in performance that makes tackling bigger and more difficult problems feasible. In addition, GPUs offer unparallel power efficiency in terms of TFlop/Watt, a feature that is becoming all the more relevant in a world of dwindling natural resources.

The only problem is that GPU architectures are a breed apart from traditional multicore CPUs. Thousands of cores, coupled with complex hierarchies of memory subsystems, constitute their efficient programming a challenge requiring specialized software platforms.

In this chapter we cover one of the most mature and feature-rich software ...

Get Multicore and GPU Programming now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.