#1889: Regression in concurrency performance from ghc 6.6 to 6.8
-----------------------------------------------+----------------------------
Reporter: dons | Owner: simonmar
Type: run-time performance bug | Status: closed
Priority: high | Milestone: 6.8.3
Component: Runtime System | Version: 6.8.1
Severity: normal | Resolution: wontfix
Keywords: threads, concurrency, performance | Difficulty: Unknown
Testcase: | Architecture: Multiple
Os: Multiple |
-----------------------------------------------+----------------------------
Changes (by simonmar):
* status: new => closed
* resolution: => wontfix
Comment:
I've investigated this further. The small regression in performance is
due to the introduction of a write barrier on `MVar`s. There isn't much
we can do to improve this, however benchmarks that have a lot of `MVar`s
(especially a lot of inactive `MVar`s) will see a big improvement in
6.8.2. This particular benchmark is not one of the winners, sadly, but
the system scales to much larger numbers of `MVar`s better now.
Indeed, this benchmark will regress again in 6.10 with the introduction of
a write barrier on TSOs, but there will be bigger winners due to that
change too.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/1889#comment:8>
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
