Submit a News Story

News

AMD OpenCL optimization case study for Support Vector Machine training

AMD offers a new Optimization case study. Examining key kernels utilized in a Quadratic Programming solver to optimize the evaluation of the Radial Basis Function SVM, results in improving performance by a factor of 5 compared to naive code running on an AMD Radeon™ HD 5870 GPU. Support Vector Machines (SVMs) are a widely used binary classification technique used to label data as belonging to one of two categories.

safety