Writing own-vector algorithms in OpenJDK* for faster performance

In this paper, we discuss insights into Vector API, which is being developed as part of OpenJDK* under Project Panama. First, we’ll go over some Vector API fundamentals, basic functionalities, and tips. We’ll then show you some code samples of vector algorithms for standard Machine Learning routines and financial benchmarks, and go over some ways to increase performance. These examples should give you some guidelines and best practices for vector programming in Java*, to help you to write successful vector versions of your own compute-intensive algorithms

Download complete paper (PDF)
 

For more complete information about compiler optimizations, see our Optimization Notice.