For those curious about how the system performance is impacted by applying compiler optimizations to the AMD A10-6800K "Richland" APU, here's some benchmarks of GCC 4.8.1 on Ubuntu Linux.
Most often when carrying out compiler benchmarks at Phoronix it's done with Intel CPUs since there's a whole lot more Intel processors around here than AMD hardware, since Intel proactively sends over a whole lot more samples than AMD has in recent times. With having bought the AMD A10-6800K this week, over the weekend I ran some GCC 4.8.1 compiler benchmarks out of curiosity.
GCC vs. LLVM/Clang benchmarks on this Richland desktop APU will be coming in the next few days, but for this Sunday article are just some GCC compiler tuning benchmarks when trying different -march= CFLAGS/CXXFLAGS and running a variety of source-based benchmarks via the Phoronix Test Suite.
The results aren't overly interesting or a huge surprise (thus just warranting this quick one page article to provide some reference numbers for those interested), but the results in full can be found in 1306303-SO-AMDA1068056 for the overclocked A10-6800K.
For some computational workloads, hitting all the instruction set extensions offered by this latest-generation 64-bit AMD APU is worthwhile but as a whole it likely won't be worth rebuilding all of your packages unless you're already running a source-based distribution.
The different -march= compiler flags are documented within the GCC online documentation.
More interesting AMD A10-6800K Linux benchmarks are forthcoming in the next few days on Phoronix. Now see the rest of the results.