[osol-discuss] Re: Re: LAMP for Solaris aka SAMP
-fast is a macro that gets expanded It will most likely overwrite everything to it's left. Correct. And the way cc and CC compilers work, if you specify an option afterwards, the last instance of the same option has precedence. At least that's what the man pages say. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Re: LAMP for Solaris aka SAMP
My previous posted results with openssl speed are void. 32 bit code was compiled with -xO3 while the 64 bit code was compiled with -xO5. I reran the tests which on average still favour 64 bit code - but to a lesser extent. One really shouldn't use -xO5 unless one has profiling data, collected from a previous sample run. At -xO5, the compiler does a lot of speculation which might or might not be correct. However, if the compiler has feedback profile data to work with, it doesn't have to speculate any more. Long story short: -xO5 is only good in tandem with a feedback profile, otherwise chances are high that the code will end up being bigger and slower. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Re: LAMP for Solaris aka SAMP
UNIX admin schrieb: Long story short: -xO5 is only good in tandem with a feedback profile, otherwise chances are high that the code will end up being bigger and slower. Short answer: The -xO5 code was faster than the -xO3 code even without profiling data. Short answer supplement: Performance tests on different architectures should have comparable optimization flags to make them meaningful. In my first benchmark run there was a discrepancy which I corrected. Daniel ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org