#1600: Optimisation: CPR the results of IO -------------------------------------+------------------------------------- Reporter: simonmar | Owner: Type: task | Status: new Priority: lowest | Milestone: Component: Compiler | Version: 6.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Runtime | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: #8598 | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by akio): I've rebased the [https://github.com/takano-akio/ghc/commits/nested-cpr branch] on top of the join points work. In nofib, allocations are generally down but the difference is not very big: {{{ -------------------------------------------------------------------------------- Program Size Allocs Runtime Elapsed TotalMem -------------------------------------------------------------------------------- bspt +0.0% +0.1% 0.007 0.007 0.0% compress2 +1.5% -0.9% 0.158 0.158 -3.7% fluid +1.6% -0.2% 0.007 0.007 0.0% infer +0.0% -1.2% 0.039 0.039 0.0% pic +0.1% -0.6% 0.005 0.005 0.0% solid +0.0% -6.6% 0.159 0.159 0.0% -------------------------------------------------------------------------------- Min -0.0% -6.6% -28.7% -28.8% -3.7% Max +1.6% +0.1% +4.3% +4.3% +6.7% Geometric Mean +0.1% -0.1% -14.7% -14.5% +0.0% }}} I'll look more carefully to see why the improvement is so small. I may have messed up something while rebasing. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/1600#comment:58 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler