-
Bug
-
Resolution: Fixed
-
P3
-
6
-
b28
-
generic
-
generic
This came about due to the use of -fast in the awt makefiles.
Ideally all compiler optimization flags should be isolated and done so that
other teams could take advantage of the set of flags and won't bake in
compiler specific optimization flags into their makefiles.
---
Just an idea, I was thinking we could use a
> CC_HIGHEST_OPT macro that selective Makefiles could set _OPT to,
> that way we'd have 3 possible choices:
> CC_LOWER_OPT=-xO2
> CC_HIGHER_OPT=-xO4
> CC_HIGHEST_OPT=-xO4 -xdepend ... <whatever>
> setup in j2se/make/common/Defs-solaris.gmk where all the specific
> option spellings could be.
> Then _OPT gets set to one of those by default, or a Makefile could
> _OPT=$(CC_HIGHEST_OPT)
> if it always wanted it set one way.
###@###.### 2005-2-12 22:11:05 GMT
Ideally all compiler optimization flags should be isolated and done so that
other teams could take advantage of the set of flags and won't bake in
compiler specific optimization flags into their makefiles.
---
Just an idea, I was thinking we could use a
> CC_HIGHEST_OPT macro that selective Makefiles could set _OPT to,
> that way we'd have 3 possible choices:
> CC_LOWER_OPT=-xO2
> CC_HIGHER_OPT=-xO4
> CC_HIGHEST_OPT=-xO4 -xdepend ... <whatever>
> setup in j2se/make/common/Defs-solaris.gmk where all the specific
> option spellings could be.
> Then _OPT gets set to one of those by default, or a Makefile could
> _OPT=$(CC_HIGHEST_OPT)
> if it always wanted it set one way.
###@###.### 2005-2-12 22:11:05 GMT
- relates to
-
JDK-6249015 REG: performance regression for 2D VIS loops in mustang-b28
-
- Closed
-