rbdlsim/3rdparty/vectorial
Martin Felis e8553fff6c Initial works on render commands 2020-10-17 23:20:38 +02:00
..
bench Initial works on render commands 2020-10-17 23:20:38 +02:00
include/vectorial Initial works on render commands 2020-10-17 23:20:38 +02:00
spec Initial works on render commands 2020-10-17 23:20:38 +02:00
tools Initial works on render commands 2020-10-17 23:20:38 +02:00
.gitignore Initial works on render commands 2020-10-17 23:20:38 +02:00
.travis.yml Initial works on render commands 2020-10-17 23:20:38 +02:00
LICENSE Initial works on render commands 2020-10-17 23:20:38 +02:00
Makefile Initial works on render commands 2020-10-17 23:20:38 +02:00
README Initial works on render commands 2020-10-17 23:20:38 +02:00
vectorial.sln Initial works on render commands 2020-10-17 23:20:38 +02:00
vectorial.vcproj Initial works on render commands 2020-10-17 23:20:38 +02:00
vectorialbenchmark.vcproj Initial works on render commands 2020-10-17 23:20:38 +02:00

README

    Vectorial - vector math library



  Motivation

    I couldn't find an open source math library that was usable and
    supported simd - especially the ARM NEON variant.


  Features

    Supports NEON, SSE, scalar and generic gcc vector extension.
    Most basic vector and matrix math is available, but not quite
    yet full featured.


  Design

    Vectorial consists of two main parts, pure-C wrapper around
    platform-specific vector instructions in the simd*.h files
    and C++ classes for common uses, the vec*.h and mat*.h

    The config.h autodetects approriate vector instructions to use.

    The platform-specific support is done with intrisincs only,
    allowing the compiler to have a full view of the code, hopefully
    resulting in better optimizations especially with reordering etc.


  Installation / Usage

    Add vectorial/include to your include path

    #include "vectorial/simd4f.h"  
    for C-only simd wrapper, using it looks like this:
      simd4f v = simd4f_normalize( simd4f_add( simd4f_create(1,2,3,4), y) );
      float z = simd4f_get_z(v);

    #include "vectorial/vectorial.h"
    for C++ classes. They reside in vectorial namespace, you might
    want to alias them to your own namespace
      namespace myproject {
        using namespace ::vectorial;
        // if you like different name: typedef vec3f Vector3;
      }
      using myproject::vec4f;
      
      vec4f v = normalize( vec4f(1,2,3,4) + y );
      float z = v.z();


  License

    2-clause BSD. See LICENSE