[Webkit-unassigned] [Bug 133136] Latest emscripten life benchmark is 4x slower

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=133136


Filip Pizlo <fpizlo at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|webkit-unassigned at lists.web |fpizlo at apple.com
                   |kit.org                     |




--- Comment #2 from Filip Pizlo <fpizlo at apple.com>  2014-05-23 15:53:46 PST ---
Created an attachment (id=231997)
 --> (https://bugs.webkit.org/attachment.cgi?id=231997&action=review)
it begins

I'm going to make a comprehensive fix here.  In most places where we would speculate integer or number, we can accept booleans also.  We should of course be smart about it - like, if we have it on good authority that the input to a math operation is *only* integers then we shouldn't emit code to allow booleans.  I think that also we need to tweak the slow path counting in the baseline JIT - the baseline will think that booleans require taking slow path even though they basically don't.  It'll be interesting to figure out what to do about that.

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