diff options
author | Simon Peyton Jones <simonpj@microsoft.com> | 2017-02-26 21:53:31 -0500 |
---|---|---|
committer | David Feuer <David.Feuer@gmail.com> | 2017-02-26 21:54:30 -0500 |
commit | 4f38fa100091152e6497db384af1fecd628e11e5 (patch) | |
tree | 0c2ab15e62ea6d9023bf6a064e807ca7b16c1f74 /docs | |
parent | e4188b538bfc879b201d416cf1d68ff7072c577f (diff) | |
download | haskell-4f38fa100091152e6497db384af1fecd628e11e5.tar.gz |
Add -fspec-constr-keen
I discovered that the dramatic imprvoement in perf/should_run/T9339
with the introduction of join points was really rather a fluke, and
very fragile.
The real problem (see Note [Making SpecConstr keener]) is that
SpecConstr wasn't specialising a function even though it was applied
to a freshly-allocated constructor. The paper describes plausible
reasons for this, but I think it may well be better to be a bit more
aggressive.
So this patch add -fspec-constr-keen, which makes SpecConstr a bit
keener to specialise, by ignoring whether or not the argument
corresponding to a call pattern is scrutinised in the function body.
Now the gains in T9339 should be robust; and it might even be a
better default.
I'd be interested in what happens if we switched on -fspec-constr-keen
with -O2.
Reviewers: austin, bgamari
Reviewed By: bgamari
Subscribers: thomie
Differential Revision: https://phabricator.haskell.org/D3186
Diffstat (limited to 'docs')
-rw-r--r-- | docs/users_guide/using-optimisation.rst | 12 |
1 files changed, 11 insertions, 1 deletions
diff --git a/docs/users_guide/using-optimisation.rst b/docs/users_guide/using-optimisation.rst index 9436832611..e56c47312c 100644 --- a/docs/users_guide/using-optimisation.rst +++ b/docs/users_guide/using-optimisation.rst @@ -522,7 +522,7 @@ list. Turn on call-pattern specialisation; see `Call-pattern specialisation for Haskell programs - <http://research.microsoft.com/en-us/um/people/simonpj/papers/spec-constr/index.htm>`__. + <https://www.microsoft.com/en-us/research/publication/system-f-with-type-equality-coercions-2/>`__. This optimisation specializes recursive functions according to their argument "shapes". This is best explained by example so consider: :: @@ -580,6 +580,16 @@ list. body directly, allowing heavy specialisation over the recursive cases. +.. ghc-flag:: -fspec-constr-keen + + :default: off + + If this flag is on, call-patten specialision will specialise a call + ``(f (Just x))`` with an explicit constructor agument, even if the argument + is not scrutinised in the body of the function. This is sometimes + beneficial; e.g. the argument might be given to some other function + that can itself be specialised. + .. ghc-flag:: -fspec-constr-count=<n> :default: 3 |