NVIDIA Developer Zone

CUDA Math Library

The CUDA Math library is an industry proven, highly accurate collection of standard mathematical functions.  Available to any CUDA C or CUDA C++ application simply by adding “#include math.h” in your source code, the CUDA Math library ensures that your application benefits from high performance math routines optimized for every NVIDIA GPU architecture.

Numerically intensive GPU-accelerated applications and libraries, including all of the CUDA libraries available from NVIDIA, rely on the CUDA Math library to deliver breakthrough results.

Key Features

 
  • Complete support for all C99 standard float and double math functions
  • IEEE-754 accurate for float, double, and all rounding modes
  • Extended Trigonometry and Exponential Functions
    • cospi, sincos, sinpi, exp10
  • Additional Inverse Error Functions
    • erfinv, erfcinv
  • Optimized Reciprocal Functions
    • rsqrt, rcbrt
  • Floating Point Data Attributes
    • signbit, isfinite, isinf, isnan
  • *New in CUDA 4.1* Bessel Functions
    • j0,j1,jn,y0,y1,yn
   

Availability

The CUDA Math library is freely available as part of the CUDA Toolkit at www.nvidia.com/getcuda.
For more information on the CUDA Math library and other CUDA math libraries: