Vector DSP Libraries

Continuum Vector DSP Libraries

 

The industry-standard Continuum Vector DSP library is designed to facilitate the programming and portability of signal processing code between numerous hardware platforms. Over 260 subroutines make up the current library, with functions that do FFTs, convolution, filtering, function generation, and basic math. The library is based on, and built from the standards set by Floating Point Systems, Inc.; the Society of Exploration Geophysicists - Technical Standards Committee; and Quantitative Technology Corporation.

 

The current Vector processing library is made up of both vector operations and matrix operations. These operations also include functions to perform single floating point precision, double precision, interleaved complex, split complex, and integer mathematics. The following four subsections split the Vector functions into four parts:

  • Interleaved complex functions;
  • Split complex functions;
  • Real functions;
  • Integer functions;

 

We provide highly optimised Continuum Vector DSP libraries for:

  • ARM A53, A57, and A72
  • PPC T2080
  • PPC T2081
  • PPC T4240
  • Intel AVX512
  • Intel Phi
  • Intel AVX2
  • Intel AVX
  • Intel SSE

 

Full reference manual documentation can be found here: 

Continuum Vector Reference Manual

For a spreadsheet of benchmark information please click on the following link:

Vector Benchmark Data on Intel Haswell.

 

 


Print   Email