Evolving AVX512 Parallel C Code Using GP

Abstract

Using 512 bit Advanced Vector Extensions, previous development history and Intel documentation, BNF grammar based genetic improvement automatically ports RNAfold to AVX, giving up to a 1.77 fold speed up. The evolved code pull request is an accepted GI software maintenance update to bioinformatics package ViennaRNA.

Genetic Improvement of LLVM Intermediate Representation

“…We have demonstrated genetic improvement of real world GPU applications [18,19,20,21], including BarraCUDA [22], the first GI code to be accepted into actual use [23]. At EuroGP'19 [24], we showed GI could also speed up parallel CPU code, this time Intel AVX vector instructions were optimised. The resulting GIed RNAfold [25] was accepted into production and like the GI version of BarraCUDA has been downloaded many thousands of times (for example [26]).…”