Benchmarking The New Optimization Level In GCC 4.8
Written by Michael Larabel in Software on 12 February 2013 at 02:02 AM EST. Page 3 of 4. 5 Comments.
The -Og performance with Himeno was in line with -O1, which was more than twice the speed of -O0. -Ofast and -O3 were nearly twice as fast as -Og but at the cost of reduced debuggability for the compiled binaries.
Building PHP as a test to see the compile time when using -Og, it was noticeably quicker than -O1 and the higher modes but at a noticeably longer time than -O0.
The C-Ray benchmark with -Og led to performance levels in line with -O1 and -O2, which were much better off than the -Os and -O0 levels.