NVIDIA GPU's have extremely high numerical throughput, both double and single precision, and applications with high arithmetic density can enjoy amazing GPU acceleration. To help you leverage this capability there are many options from technical application solutions to numerical libraries. See some of these options below:

	Matlab by Mathworks, has native support for CUDA in the Parallel Computing Toolbox. Enjoy GPU acceleration without having to write any CUDA Kernels.
	Mathematicia by Wolfram is a comprehensive technical computing solution, enabling complex computational applications to be build, and with native GPU acceleration.
	Accelereyes libjacket is a comprehensive GPU Function library including functions for math, signal processing , statistics and much more.
	IMSL Fortran Numerical Library is a comprehensive set of mathematical and statistical functions that developers can embed into their Fortran software applications.
	LabView enables engineers and scientists to create applications using a powerful high level programming language and advanced tools, and now features GPU acceleration support.

 
Looking for expert advice on how to best use the numerical capabilities of your GPU?
Try reviewing the documentation and educational materials below or get in touch with industry experts and NVIDIA engineers on the CUDA Developer forums