I am running some floating point benchmarks to determine the overhead in
changing some stuff around in how gcc/glibc do floating point to address
the issues here:
Right now, I am trying to measure the overhead of running a floating
point benchmark with shared-libgcc vs. static-libgcc. Can I have access
to a real ARM machine running Fedora? I am not sure if QEMU is totally
accurate in these benchmarks.