I tried this option on several other compilers.
On Windows, it gives a warning that the option is not recognized. But
the compile does complete.
On AIX, it gives a warning for solver.o:
1500-036: (I) The NOSTRICT option (default at OPT(3)) has the
potential to alter the semantics of a program. Please refer to
documentation on the STRICT/NOSTRICT option for more information.
On Linux and Mac OS, we use gcc so it should be ok.
Tom
On 1/8/2010 4:26 PM, Danek Duvall wrote:
Would anyone object to the following patch:
--- a/src/setup.py
+++ b/src/setup.py
@@ -777,7 +777,7 @@ class dist_func(_bdist):
# These are set to real values based on the platform, down below
-compile_args = None
+compile_args = [ "-O3" ]
link_args = None
ext_modules = [
Extension(
It adds a "-O3" flag to the end of every compile. It will require that you
use SS12 or newer (or gcc) as your compiler, which is probably okay. But
it gives us a small speedup in action parsing, and a large speedup in
solver performance on Niagara.
On x86 and sparc64, the solver speeds up by about 3x; on N2, the solver
speeds up by about 2.6x. (Average of three runs of run.py -o t_solver.)
On x86, action parsing speeds up by about 7.6%; on sparc64, by 3.6%; on N2,
by 3.2%. (Average of the three runs of action creation in actionbench.)
Danek
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
