#2271: floor, ceiling, round :: Double -> Int are awesomely slow
------------------------------------------+---------------------------------
Reporter: dons | Owner: daniel.is.fischer
Type: bug | Status: infoneeded
Priority: low | Milestone: 7.0.1
Component: libraries/base | Version: 7.1
Keywords: performance, math, double | Testcase:
Blockedby: | Difficulty: Unknown
Os: Unknown/Multiple | Blocking:
Architecture: Unknown/Multiple | Failure: None/Unknown
------------------------------------------+---------------------------------
Changes (by daniel.is.fischer):
* status: patch => infoneeded
Comment:
Great, I just was notified that there's a bug in rintf for some versions
of glibc (apparently only on x86_64):
http://sources.redhat.com/bugzilla/show_bug.cgi?id=10089
And thus round :: Float -> Int gets some values wrong there.
{{{
Cround/Int
*** Failed! Falsifiable (after 2333 tests):
2788790.8
Failure {usedSeed = 678847334 2020505879, usedSize = 2332, reason =
"Falsifiable", labels = []}
Now the funny part is that this seems to be rintf() bug in glibc
(only x86_64), see
http://sources.redhat.com/bugzilla/show_bug.cgi?id=10089
(I've write a small C program to check the output for this number and it
does return the wrong result, i.e. 2788790).
So I thought that I'll let you know about it before anyone else runs
into the same issue. All the other tests passed just fine, btw.
}}}
I'll see whether I can cook up a fast bit-fiddler in C then.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/2271#comment:18>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
