Vector Accelerator(patent pending) is designed for cases where user simply wants to process multiple samples of data and utilize SIMD(AVX2 or AVX512) and multi-thread to gain a x200 magnitude increase in speed performance. Writing Multi-thread safe code for large projects can be challenging. This library transforms your calculations into versions which can be safely executed in parallel even when the original code is not. Minimal code changes made to the original serial code and large projects can be accelerated.
Very large computations are possible. Efficient memory use and code folding techniques.
Take full advantage of AVX2 or AVX512 native CPU vectorization
Output function is safe to execute in parallel even when the original code is not.
No template expressions. Unique active type allows to reuse in-memory objects.
Automatically tested on multiple platforms/compilers using rich set of test cases
Keeps original code readable and easy to maintain