GCC 4.9 Compiler Optimization Benchmarks For Faster Binaries

[hubicka]

Except for the unit tests me neither, it is on my TODO.

Other half of answer ended up in QUOTE: TALOS seems to do that (I am playing with it now). Dromaeo indeed needs one mouse click but there is probably way how to script that.

[Michael]

These things are extremely useful Michael (I hope you are sending bugreports for the regressions with O3 being slower than O2).

As I've explained before, I generally don't have the time/resources to do bug reports when writing 10+ articles per day and working on an equal number or greater tests each day by myself, besides software development and other work. Can't really afford to spend any time on bug reports when it's hard enough making a living off the Phoronix.com site as is but besides by the time I write a bug report, the chances are the system has already changed around -- either the hardware components, or the distro completely wiped and doing other sorts of tests, etc. So it doesn't work out for testing fixes, etc.

Would it be possible to benchmark -Ofast -march=native -flto ?

Since -Ofast enables -O3 + -ffast-math I would expect it to be the most aggresive "easy to activate" optimization level.

Can add it to the list for next time I do such compiler tests...

[0xBADCODE]

Curious about -fwhole-program

Curious how the same setup performs with -fwhole-program. Also, it can be nice to compare binary sizes. While it really matters only in embedded, test compile with -flto -fwhole-program made large test C++ binaries about 25% smaller with no other efforts or obvious loses.