Comment #1 on issue 2038 by [email protected]: Implement optimization for
Math.floor(a/b) for non ARM architectures
http://code.google.com/p/v8/issues/detail?id=2038
A proposal for x86/x64 implementation of Math.floor(a/b).
Only handles when "a" is int32 and "b" is int32 constant.
Diff against Patch Set 2 of the above review.
No review yet because of dependency.
Currently implemented by imul (not fpmul).
x86 and x64 algorithm differs a bit.
Attachments:
MathFloorOfDiv.x86.patch 16.3 KB
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
