Vector Library Performance Test¶
Section author: Luke Frisken <l.frisken@gmail.com>
The purpose of this performance test was to review several methods/libraries available for vector/matrix math operations.
The source code for the project can be found at https://github.com/kellpossible/java-vector-math-performance-test
- Apache commons-math
- Pythagoras Library
- Yeppp! Library
- LWJGL OpenCL (TODO)
Vector classes using double precision floating point numbers were used. Where classes were missing, a plain java alternative was implemented.
Performance Test Results¶
| Benchmark | Samples | Score | Error | units |
|---|---|---|---|---|
| testApacheCommons | 20 | 1.916 | ± 0.428 | ops/s |
| testPythagoras | 20 | 7.572 | ± 0.355 | ops/s |
| testYepp | 20 | 0.205 | ± 0.001 | ops/s |
Review of Libraries¶
Apache commons-math¶
- lots of other classes we may want to use
- bloated
- ~ performance is okay but not the best.
Pythagoras¶
The Pythagoras has the best performance. It is also small
- very small
- performance
- no unit tests for double class
Yeppp!¶
Yeppp doesn’t have the best performance for our application, and we’d have alter the design to cater for performance to get any kind of benefit.
- small
- popular
- poor performance in java and for our application
c++ - Performance with Yeppp! is slower than native implementation - Stack Overflow > Yeppp! is optimized for processing arrays of 100+ elements. > It is not efficient on small arrays (like length-3 array in your example) > due to limited ability to use SIMD and overheads of function call, dynamic > dispatching, and parameter checks.