#7258: Compiling DynFlags is jolly slow -------------------------------------+------------------------------------- Reporter: simonpj | Owner: simonpj Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.6.1 Resolution: | Keywords: deriving-perf Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by tdammers): Replying to [comment:48 simonpj]:
Is the labelling in the graph after comment:46 correct? read-appl looks worst by a long way, which wasn't true bef.re Ditto geline-appl seems worse than getline (which is invisible).
Yes, it is correct; the difference between this new graph and the previous one is that the new one uses `-O2`, while the old one doesn't use any optimizations. The profiler breakdown looks radically different between the optimized and unoptimized versions, for *all* examples - RegAlloc- linear and ppr, which are the most serious offenders in the unoptimized run, account for less than 1% of execution time in the optimized case. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/7258#comment:50 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler