Mark Allums wrote:
I guess someone will need to hook me up with the test codecs, if we
want to be consistent with previous work. However, I can use any
codecs available if all we want to do is compare gcc 4.0.3 to gcc 4.4.0.
Consistency with previous work isn't the goal here. Since you haven't
made any code changes, you should just take a Rockbox build of the same
SVN revision as compiled on the GCC we recommend, and perform the same
benchmark, against the same audio files, with a Rockbox compiled with 4.4.0.
Attempting to compare to the benchmarks on that page will be useless
unless you compile for the same target, and the same SVN revision.
The goal is to identify differences in GCC optimization, not differences
cross-revision.
