#14644: Improve cmm/assembly for pattern matches with two constants. -------------------------------------+------------------------------------- Reporter: AndreasK | Owner: AndreasK Type: task | Status: patch Priority: normal | Milestone: Component: Compiler | Version: 8.2.2 (CodeGen) | Keywords: Codegen, CMM, Resolution: | Patterns, Pattern Matching Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Phab:D4294 Wiki Page: | -------------------------------------+------------------------------------- Comment (by AndreasK): Replying to [comment:9 simonpj]:
Didn't you say "I ran a simple Benchmark and the results seem almost too good with >10% gains in almost all cases.". But now you say "mostly no change; one gets 5-9%". Were the earlier measurements wrong?
No these are right. But the simple benchmark was running the code I posted above in [https://ghc.haskell.org/trac/ghc/ticket/14644?replyto=9#comment:5 comment 5]. All cases referred to the pos/neg/.. branches in that case. Nofib seems to be in a permanent state of breakdown on windows so I stopped using it in my initial benchmarks. And it was the [https://ghc.haskell.org/trac/ghc/ticket/14656 right] [https://ghc.haskell.org/trac/ghc/ticket/14655 thing] [https://ghc.haskell.org/trac/ghc/ticket/14654 to do ] as it took some work to get it running. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/14644#comment:10 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler