On 07/03/2012 08:19 PM, Larry Evans wrote:
Anyone know why, in this particular case, -O3 is faster than with no optimization?
Without optimizations Execution times (seconds) phase setup : 0.01 phase parsing : 0.10 phase lang. deferred : 0.54 phase cgraph : 3.50 phase generate : 4.04 |name lookup : 0.08 |overload resolution : 0.30 With optimizations Execution times (seconds) phase setup : 0.01 phase parsing : 0.09 phase lang. deferred : 0.52 phase cgraph : 0.40 phase generate : 0.92 |name lookup : 0.03 |overload resolution : 0.34 Optimizations done by the frontend resulted in much less symbols to compute the call graph for.