#15998: GHC.Event.Thread.eventManager has a lot of indirections -------------------------------------+------------------------------------- Reporter: chessai | Owner: (none) Type: feature request | Status: new Priority: normal | Milestone: 8.6.3 Component: libraries/base | Version: 8.6.2 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by carter): Chessai: have you done any benchmarks / or experiments for this thesis? I do have the slightly worry that this sort of optimization/unboxing might back fire on these high concurrency data structures. Or have minimal impact. The workloads which have these be the bottle neck are gonna be exactly those which have the data structure already in cache I believe. These sorts of representation changes really need to be motivated by clear benchmarkable wins. One spot that’s related which I believe would benefit from some data structure optimization is the priority heap queue used by the timer manager. It’s a pretty simple implementation and for web facing applications I think timers are used quite heavily. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15998#comment:1 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler