Vector processing

From Computer History Wiki
Jump to: navigation, search

Vector processing support refers to the ability of CPUs to provide direct support in hardware for handling vectors (in the same way that floating point support refers to handling computations using floating point quantities directly, rather than doing those computations in software, with more elementary operations).

Built in support for vector operations results in increased performance, especially in pipelined CPUs, where elements in the vector can be fed through the pipeline faster by a vector operation than they could be by individual instructions.

It first appeared in the very largest supercomputers, notably the STAR-100 and CRAY-1, and worked their way down though ordinary mainframes, etc.