benchmark thread-safe hashconsing #486
Description
See #482 :
The thing is that I'm afraid this one is going to make performance worst a lot. I'd like to benchmark it first before we start using it on Smt.ml. If the performances are fine, then I don't mind using it directly. Otherwise, I have some ideas on how we could improve the situation (one hashcons cache per type of expression, or simply use a lock-free hash table). I'll open an issue about this one too.