#9304: Floating point woes; Different behavior on 32- vs 64-bit x86 -------------------------------------+------------------------------------ Reporter: lerkok | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.8.3 Resolution: | Keywords: floating point Operating System: Unknown/Multiple | Architecture: Unknown/Multiple Type of failure: None/Unknown | Difficulty: Unknown Test Case: | Blocked By: 9276 Blocking: | Related Tickets: -------------------------------------+------------------------------------ Comment (by rwbarton): Replying to [comment:26 simonpj]:
However I don't understand Reid's comment that "the optimiser is constant-folding the multiplication". Which optimiser?
I mean the rule {{{ primOpRules nm DoubleMulOp = mkPrimOpRule nm 2 [ binaryLit (doubleOp2 (*)), ... ] }}} which we can see firing by compiling with `-ddump-rule-firings`: {{{ ... Rule fired: *## ... }}} The constant folder represents literal Double values with Rational, but (unless excess precision is enabled) truncates the result of each operation by converting to Double and back, so it should match the answer obtained by using 64-bit Double throughout, as it does in this case. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/9304#comment:28 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler