#5113: Huge performance regression of 7.0.2, 7.0.3 and HEAD over 7.0.1 and 6.12
(MonoLocalBinds)
--------------------------------------------+-------------------------------
Reporter: daniel.is.fischer | Owner:
Type: bug | Status: new
Priority: normal | Milestone:
Component: Compiler | Version: 7.0.3
Keywords: performance, MonoLocalBinds | Testcase:
Blockedby: | Difficulty:
Os: Linux | Blocking:
Architecture: x86 | Failure: Runtime
performance bug
--------------------------------------------+-------------------------------
Comment(by daniel.is.fischer):
Yes, it is polymorphic, but it's a local binding inside a runST, only ever
used at the one specific type coming from the enclosing ST-action. Up to
and including 7.0.1, it was automatically inlined, resulting in
{{{
case GHC.Prim.readIntArray#
@ s_aKv marr#_aLG sc1_sSa sc_sS9
of _ { (# s2#1_aPz, e#_aPA #) ->
case GHC.Prim.writeIntArray#
@ s_aKv
marr#_aLG
sc1_sSa
(GHC.Prim.+# e#_aPA 1)
s2#1_aPz
of s2#2_aQ2 { __DEFAULT ->
}}}
at the appropriate places. 7.0.2, 7.0.3 and HEAD (29.3. - I've not yet
built a newer HEAD to see whether it changed) produce the polymorphic
worker that's then called.
As for specific modules, I've not yet managed to produce a small example
exhibiting exactly that, but for the small `Test.hs` module, 7.0.3
produces bad (polymorphic) core while HEAD (and 7.0.1) do fine. However,
for the `display` bundle (the interesting module is `DispFloat`), HEAD
also produces the polymorphic `$winc`.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5113#comment:3>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
