[Webkit-unassigned] [Bug 65438] JSC GC is far too conservative about growing the heap size, particularly on desktop platforms

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Mon Aug 1 17:21:45 PDT 2011


https://bugs.webkit.org/show_bug.cgi?id=65438





--- Comment #10 from Gavin Barraclough <barraclough at apple.com>  2011-08-01 17:21:45 PST ---
Maciej had some comments on irc, pasting them here - I think his key suggestion is that LARGE_HEAP should be the default, and platforms should opt in to SMALL_HEAP.

[12:05am] othermaciej: olliej_: do we have a "low memory" ifdef of some kind?
[12:05am] othermaciej: would be much better to trigger small vs large heap off of that instead of MAC
[12:05am] othermaciej: it would also be useful to know the memory cost of this patch on something like membuster
[12:05am] gbarra: Given the performance constraints of our current GC, the idea that if a page rapidly generates 16Mb of objects then our heap usage may go up by 16Mb seems reasonable to me.
[12:05am] gbarra: othermaciej: *nod.  I can't think of one.
[12:06am] othermaciej: in fact LARGE_HEAP is presumably the normal case so it should be SMALL_HEAP that should be the specialized case, since it is dependent on a temporary limitation of some platforms
[12:07am] gbarra: othermaciej: seems reasonable.  I think Filips thought process was that there are more embedded CPUs than there are desktop CPUs, but I think you have a good point.
[12:08am] othermaciej: it's really not CPU-based, it is platform based
[12:08am] gbarra: *nod.
[12:08am] othermaciej: and most target platforms are not embedded
[12:08am] gbarra: true.

-- 
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