[Webkit-unassigned] [Bug 88314] Improve Math.round and Math.floor intrinsic

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=88314


Filip Pizlo <fpizlo at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #145729|review?                     |review+
               Flag|                            |




--- Comment #4 from Filip Pizlo <fpizlo at apple.com>  2012-06-05 20:47:07 PST ---
(From update of attachment 145729)
View in context: https://bugs.webkit.org/attachment.cgi?id=145729&action=review

This looks right.  I'm curious what your thoughts are about the comparing against neg zero.  Maybe this code would work the same if you just used zero instead of neg zero, which would be faster, since regular zero is easier to load in SSE?

Not marking cq+ just because I want to give you an opportunity to respond.  But I think the code is correct so if there's nothing that needs changing then I'm fine with this being landed as-is.

> Source/JavaScriptCore/jit/ThunkGenerators.cpp:197
> +        doubleResult.append(jit.branchDouble(MacroAssembler::DoubleEqual, SpecializedThunkJIT::fpRegT0, SpecializedThunkJIT::fpRegT1));

Is this right?  Double equality has strange handling of zero.  0 == -0 for example.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.