Re: [GHC] #8763: forM_ [1..N] does not get fused (allocates 50% more)

#8763: forM_ [1..N] does not get fused (allocates 50% more) -------------------------------------+------------------------------------- Reporter: nh2 | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.8.1 Component: Compiler | Version: 7.6.3 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Runtime | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: #7206 | Differential Rev(s): Phab:D5131 Wiki Page: | -------------------------------------+------------------------------------- Comment (by Krzysztof Gogolewski ): In [changeset:"e655aac18c5b240f27fcaf26317ad87be5ce8b96/ghc" e655aac1/ghc]: {{{ #!CommitTicketReference repository="ghc" revision="e655aac18c5b240f27fcaf26317ad87be5ce8b96" Make sure forM_ and related functions fuse cleanly Summary: It was revealed in #8763 that it's hard to come up with a list fusion helper for `efdtIntFB` that doesn't duplicated occurrences of `c`, which is crucial in guaranteeing that it is inlined. Not inlining `c` led to spoiled join points, in turn leading to unnecessary heap allocation. This patch tackles the problem from a different angle: Fixing all consumers instead of the less often used producer `efdtIntFB` by inserting an INLINE pragma in the appropriate places. See https://ghc.haskell.org/trac/ghc/ticket/8763#comment:76 and the new Note [List fusion and continuations in 'c']. A quick run of NoFib revealed no regression or improvements whatsoever. Reviewers: hvr, bgamari, simonpj Reviewed By: simonpj Subscribers: simonpj, rwbarton, carter GHC Trac Issues: #8763 Differential Revision: https://phabricator.haskell.org/D5131 }}} -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/8763#comment:85 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler