[Webkit-unassigned] [Bug 20764] Proposed byte-code optimization framework
bugzilla-daemon at webkit.org
bugzilla-daemon at webkit.org
Fri Sep 12 05:27:53 PDT 2008
https://bugs.webkit.org/show_bug.cgi?id=20764
------- Comment #7 from zherczeg at inf.u-szeged.hu 2008-09-12 05:27 PDT -------
Hi,
We have found some good peephole optimizations in SunSpider.
1) Optimizing negates:
[ 434] negate tr132, lr2
[ 437] negate tr133, lr2
[ 440] negate tr134, lr2
(sometimes a move operation is inserted between the negates)
Can be changed to:
[ 434] negate tr132, lr2
[ 437] op_mov tr133, tr132
[ 440] op_mov tr134, tr132
2) Mov folding:
a)
[ 6] mov tr36, tr0
[ 9] mov tr37, tr1
[ 12] mov tr38, tr2
[ 15] mov tr39, tr3
...
b)
[ 538] mov tr132, tr3
[ 541] mov tr133, tr3
[ 544] mov tr134, tr3
[ 547] mov tr135, tr3
...
Used especially in array initialization and function call with constant
arguments.
We introduced two new instruction:
op_mov_range for the first version (implemented by a memcpy call)
op_set_range for the second version
Preliminary measurements shows 2.3% performance improve with SunSpider, but we
will see what will happen after we merge our svn with the official one in
Monday.
--
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
More information about the webkit-unassigned
mailing list