[gambit-list] macos 10.9

Bradley Lucier lucier at purdue.edu
Mon May 5 17:30:18 EDT 2014


On 05/05/2014 03:11 PM, Marc Feeley wrote:
> A simple solution would be to drop the use of -fschedule-insns2 so that clang will not raise an error if it is the C compiler that ends up being used.  That option is for performance reasons but as I recall it doesn’t improve performance by very much.

With older gcc's, it used to improve it a lot (sometimes doubling the 
speed or more).

Perhaps you should just benchmark it with and without 
-fschedule-insns2.  I just did some partial benchmarks with gcc-4.8.2 on 
my Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz and didn't notice much 
difference.

Brad



More information about the Gambit-list mailing list