Parallelism

A high level overview of SIMD architecture

Every graphics processor executes the same instruction on multiple threads concurrently, in the form of Single Instruction, Multiple Data (SIMD) processing.

The main advantage of the SIMD architecture is that significant numbers of threads can be run in parallel for a correctly structured application, and this is done with extremely high efficiency. SIMD architectures are usually capable of running many orders of magnitude more threads at once than a typical CPU.

SIMD is designed to operate on large coherent data sets and performs exceptionally well at this type of task. Algorithms that operate independently on a large coherent data set, such as graphics and image processing, are therefore well suited for this processor type.