diff options
| author | Andreas Klebinger <klebinger.andreas@gmx.at> | 2020-01-28 20:53:04 +0100 |
|---|---|---|
| committer | Marge Bot <ben+marge-bot@smart-cactus.org> | 2021-05-27 02:35:11 -0400 |
| commit | 03d69e4bb6b92ccd8238bebd9cff68da23741f49 (patch) | |
| tree | 4dbaf811ec9bda4c9544530f63509ad8c7062bc3 /docs/users_guide | |
| parent | 9faafb0aaff04e86a58b9e108f84618b12f2057c (diff) | |
| download | haskell-03d69e4bb6b92ccd8238bebd9cff68da23741f49.tar.gz | |
Enable strict dicts by default at -O2.
In the common case this is a straight performance win
at a compile time cost so we enable it at -O2.
In rare cases it can lead to compile time regressions
because of changed inlining behaviour. Which can very
rarely also affect runtime performance.
Increasing the inlining threshold can help to avoid this
which is documented in the user guide.
In terms of measured results this reduced instructions executed
for nofib by 1%.
However for some cases (e.g. Cabal) enabling this
by default increases compile time by 2-3% so we enable it only
at -O2 where it's clear that a user is willing to trade compile
time for runtime.
Most of the testsuite runs without -O2 so there are few
perf changes.
Increases:
T12545/T18698: We perform more WW work because dicts are now treated strict.
T9198: Also some more work because functions are now subject to W/W
Decreases:
T14697: Compiling empty modules. Probably because of changes inside ghc.
T9203: I can't reproduce this improvement locally. Might be spurious.
-------------------------
Metric Decrease:
T12227
T14697
T9203
Metric Increase:
T9198
T12545
T18698a
T18698b
-------------------------
Diffstat (limited to 'docs/users_guide')
| -rw-r--r-- | docs/users_guide/using-optimisation.rst | 11 |
1 files changed, 10 insertions, 1 deletions
diff --git a/docs/users_guide/using-optimisation.rst b/docs/users_guide/using-optimisation.rst index 3fcac45e11..31d3a71431 100644 --- a/docs/users_guide/using-optimisation.rst +++ b/docs/users_guide/using-optimisation.rst @@ -343,7 +343,7 @@ by saying ``-fno-wombat``. seem cheap to the optimiser. .. ghc-flag:: -fdicts-strict - :shortdesc: Make dictionaries strict + :shortdesc: Make dictionaries strict. Implied by :ghc-flag:`-O2`. :type: dynamic :reverse: -fno-dicts-strict :category: @@ -352,6 +352,15 @@ by saying ``-fno-wombat``. Make dictionaries strict. + This enables WW to fire on dictionary constraints which usually results + in better runtime. In niche cases it can lead to significant compile time + regressions because of changed inlining behaviour. Rarely this can also affect + runtime negatively. + + If enabling this flag leads to regressions try increasing the unfolding + threshold using :ghc-flag:`-funfolding-use-threshold=⟨n⟩` by a modest amount (~30) + as this is likely a result of a known limitation described in `#18421`. + .. ghc-flag:: -fdmd-tx-dict-sel :shortdesc: *(deprecated)* Use a special demand transformer for dictionary selectors. :type: dynamic |