[llvm] [KeyInstr] MDNodeKeyImpl<DILocation> skip zero values for hash (PR #143357)

[Nikita Popov via llvm-commits]

https://github.com/nikic commented:

I think the main problem is that you are going from hashing 25 bytes to 34, which means you go up one hash function (> 32). I think if you packed ImplicitCode, AtomGroup and AtomRank into one uint64_t value for the hash, performance would be about unchanged from not hashing them? (Reducing AtomGroup to 60 bits.)

https://github.com/llvm/llvm-project/pull/143357