[GHC] #12765: Don't optimize coercions with -O0

#12765: Don't optimize coercions with -O0 -------------------------------------+------------------------------------- Reporter: bgamari | Owner: Type: task | Status: new Priority: normal | Milestone: 8.2.1 Component: Compiler | Version: 8.0.1 Keywords: | 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: -------------------------------------+------------------------------------- During the call today Simon and I discussed the possibility of omitting coercion optimization from the desugarer's simple optimization (`simpleOptPgm`) when compiling with `-O0`. The justification here was to perhaps speed-up unoptimized compilation times for programs like that of #12506, which has very large coercions (presumably due to #8095). The rationale was that while this first pass of coercion simplification during desugaring may reduce the size of some of these coercions, we only walk over them once more during code generation, so there's little benefit given the cost. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/12765 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler

#12765: Don't optimize coercions with -O0 -------------------------------------+------------------------------------- Reporter: bgamari | Owner: Type: task | Status: new Priority: normal | Milestone: 8.2.1 Component: Compiler | Version: 8.0.1 Resolution: | Keywords: 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 bgamari): I'm dropping a ticket here since I'm no longer convinced that this is worthwhile. Unfortunately this would require a significant amount of plumbing of the optimization level through a bunch of pure simplification functions to make this happen. Moreover, even if we forego the coercion simplification pass in `simpleOptPgm`, we still do run the simplifier even with `-O0`. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/12765#comment:1 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler
participants (1)
-
GHC