[Webkit-unassigned] [Bug 74170] DFG's interpretation of rare case profiles should be frequency-based not count-based

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=74170





--- Comment #8 from Filip Pizlo <fpizlo at apple.com>  2011-12-09 13:54:29 PST ---
(In reply to comment #7)
> (From update of attachment 118627 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=118627&action=review
> 
> r=me
> 
> > Source/JavaScriptCore/runtime/Heuristics.cpp:148
> > +    SET(likelyToTakeSlowCaseMinimumCount, 100);
> 
> It seems like this 100 limit is still problematic in the first example you gave.

Yup.  But making it lower will cause us to become overly pessimistic for code that takes many slow cases in the early part of its execution.  That's why we also need a story for realizing that the reason why recompilation was triggered, was because of some slow case in the DFG.  I'll do that in a follow-on patch, since this one is big enough already.

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