--- Comment #3 from rguenth at gcc dot gnu dot org 2006-11-10 09:21 ---
I suppose we should look at RTL loop invariant motion to hoist the shift
because we should have more precise information there.
--
rguenth at gcc dot gnu dot org changed:
What|Removed
--- Comment #2 from pinskia at gcc dot gnu dot org 2006-11-10 01:55 ---
> (it considers the shifts being of cost one)
depends on the target. On some targets (Cell), shifts with a constant shifter
is cheap (same as a normal add or subtract) while with a non constant one, it
is microcoded
--- Comment #1 from rguenth at gcc dot gnu dot org 2006-11-09 22:07 ---
If the fold transformation is disabled we can force lim to move the shifts out
of the loop by specifying --param lim-expensive=1 (it considers the shifts
being of cost one). The default of the param is 20 though.
P
