flatparse is slower than bytesmith, bets @chessai
sjakobi opened this issue · 1 comments
sjakobi commented
Quoting from https://functionalprogramming.slack.com/archives/C0433UN0M/p1615839832068800:
sjakobi: András Kovács has uploaded flatparse to Hackage! http://hackage.haskell.org/package/flatparse-0.1.0.0
chessai: Willing to bet money it's still slower than bytesmith
...
sjakobi: BTW, how much money, and can I quote you on the flatparse issue tracker? :wink:
chessai: 100 USD, you can quote me
chessai: If I am right, then I will do nothing
chessai: If I am wrong, then I will just improve bytesmith
chessai: I'm not actually convinced flatparse is slower, I just think a bet makes this interesting
(CC @chessai)
AndrasKovacs commented
Added bytesmith benchmark. So far it's slower than flatparse. I included results with -fllvm from my machine in the README. You can also look at CI benchmark results which are without -fllvm. The gap narrows for s-expressions without -fllvm but not really for other benchmarks.