diff options
| author | Simon Peyton Jones <simonpj@microsoft.com> | 2022-07-15 17:47:32 +0100 |
|---|---|---|
| committer | Simon Peyton Jones <simonpj@microsoft.com> | 2022-11-10 12:20:03 +0000 |
| commit | f9f17b68b144a7ecb91395c1e987bbf4f91c0180 (patch) | |
| tree | b5e1460afc1926dd30646b6facaa26f87e031efd /libraries/base | |
| parent | 90c5abd4581b404f715e72ad55303e18d0c31d68 (diff) | |
| download | haskell-f9f17b68b144a7ecb91395c1e987bbf4f91c0180.tar.gz | |
Fire RULES in the Specialiser
The Specialiser has, for some time, fires class-op RULES in the
specialiser itself: see
Note [Specialisation modulo dictionary selectors]
This MR beefs it up a bit, so that it fires /all/ RULES in the
specialiser, not just class-op rules. See
Note [Fire rules in the specialiser]
The result is a bit more specialisation; see test
simplCore/should_compile/T21851_2
This pushed me into a bit of refactoring. I made a new data types
GHC.Core.Rules.RuleEnv, which combines
- the several source of rules (local, home-package, external)
- the orphan-module dependencies
in a single record for `getRules` to consult. That drove a bunch of
follow-on refactoring, including allowing me to remove
cr_visible_orphan_mods from the CoreReader data type.
I moved some of the RuleBase/RuleEnv stuff into GHC.Core.Rule.
The reorganisation in the Simplifier improve compile times a bit
(geom mean -0.1%), but T9961 is an outlier
Metric Decrease:
T9961
Diffstat (limited to 'libraries/base')
| -rw-r--r-- | libraries/base/GHC/Real.hs | 5 |
1 files changed, 4 insertions, 1 deletions
diff --git a/libraries/base/GHC/Real.hs b/libraries/base/GHC/Real.hs index aaff2ea3a9..4aee6ba720 100644 --- a/libraries/base/GHC/Real.hs +++ b/libraries/base/GHC/Real.hs @@ -701,11 +701,14 @@ half of y - 1 can be computed as y `quot` 2, optimising subtraction away. Note [Inlining (^) ~~~~~~~~~~~~~~~~~~ -The INLINABLE pragma allows (^) to be specialised at its call sites. +The INLINABLE [1] pragma allows (^) to be specialised at its call sites. If it is called repeatedly at the same type, that can make a huge difference, because of those constants which can be repeatedly calculated. +We don't inline until phase 1, to give a chance for the RULES +"^2/Int" etc to fire first. + Currently the fromInteger calls are not floated because we get \d1 d2 x y -> blah after the gentle round of simplification. |