Open
Description
We need to benchmark the supercompiler on some bigger example than those in the examples/
folder, to see which parts take most time. The usual suspects are substitution and homeomorphic embedding.
But before that, Mazeppa needs to be smart enough to emit reasonable code for all programs (see #2). It makes no sense to benchmark the algorithm that needs conceptual changes.
Note: all the benchmarks below were conducted on AMD Ryzen 9 5900HX.
Metadata
Assignees
Labels
No labels
Activity