[Webkit-unassigned] [Bug 85752] Truncating multiplication on integers should not OSR exit every time

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Sun May 6 16:19:18 PDT 2012


https://bugs.webkit.org/show_bug.cgi?id=85752


Gavin Barraclough <barraclough at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #140441|review?                     |review-
               Flag|                            |




--- Comment #2 from Gavin Barraclough <barraclough at apple.com>  2012-05-06 16:19:18 PST ---
(From update of attachment 140441)
View in context: https://bugs.webkit.org/attachment.cgi?id=140441&action=review

r-, because I really do think we need a comment here – but nice optimization, looks great otherwise.

> Source/JavaScriptCore/dfg/DFGGraph.h:512
> +        int32_t twoToThe17 = 131072;

Why 2^17?

A number with an absolute value less than 2^17 is representable by 17 bits.  The result of multiplying a value representable by 17 bits by 2^31 (the largest possible absolute value of a signed integer) is representable in a 48 bit mantissa (with additional representation for the sign).  Doubles support 53 bits of mantissa precision.  Put the other way around, for the result of a multiply to be constrained within a 53 bits mantissa, if I know that the absolute value of one operand is no more than 2^31 then the second operand may have a magnitude of +/- 2^22 (less than/greater than – exclusive!).

I think 2^22 is correct here, but either way this really needs a comment. :-)

Also, I'd suggest this would slightly nicer written ''int32_t  twoToThe17 = 1 << 17;".

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


More information about the webkit-unassigned mailing list