Large Language Models for Compiler Optimization

Re:Large Language Models for Compiler Optimization by haltist on Sunday September 17, 2023 @06:41PM (#37550045)

Next step is to add verification for optimized code from the LLM with an SMT solver to remove "hallucinations". If the input and output code can be verified to be equivalent then this would be a great addition to an optimization pipeline.

[ Parent | on HN ]

Re:Large Language Models for Compiler Optimization by boomanaiden154 on Sunday September 17, 2023 @07:12PM (#37550276)

There's already tooling available for using SMT to validate LLVM-IR transformations [1]. It's designed for zero false positives however, so some things might slip through the cracks.

Additionally, this work focuses on phase ordering, which produces correct code regardless of what the LLM puts out, assuming there aren't any bugs in the passes being used (which could crop up as random orderings aren't as well tested as the standard orderings in the commonly used pipelines).

1. https://users.cs.utah.edu/~regehr/alive2-pldi21.pdf

[ Parent | on HN ]

Re:Large Language Models for Compiler Optimization by haltist on Sunday September 17, 2023 @11:08PM (#37551766)

Interesting. AI code has no loops so the problem they mention in the beginning about unrolling loops to a certain depth is a non-issue.

[ Parent | on HN ]