Home    cuSVM 

A CUDA Implementation of Support Vector Classification and Regression

cuSVM is a software package for high-speed (Gaussian-kernelized) Support Vector Machine training and prediction that exploits the massively parallel processing power of Graphics Processors (GPUs).  cuSVM is written in NVIDIA's CUDA C-language GPU programming environment, includes implementations of both classification and regression, and performs SVM training (prediction) at 13-73 (22-172) times the rate of state of the art CPU software.  Moreover, cuSVM features a Matlab MEX wrapper so that users can access the GPU's power without having to do any "real" programming. 

A Short Paper Introducing cuSVM       Download