GCC 4.7 was released a few months back, but have changes in the trunk code-base -- for what will eventually become GCC 4.8 -- resulted in any major performance changes yet?
GCC 4.8 will likely not be released until early 2013, but it's worth seeing if there's any performance changes early on in case the change is a regression or other issue. Plus GCC 4.8 will be competing with what will likely be LLVM/Clang 3.2 or LLVM/Clang 3.3.
The GCC 4.8 changes page at this point only mentions a new partial redundancy elimination (PRE) optimization and a new x86_64 option concerning stack alignment. The partial redundancy elimination optimization is exposed via the -ftree-partial-pre compiker switch or is also enabled at the -O3 optimization level where it's more aggressive. Many more changes and new features will be documented in the months leading up to the GCC 4.8.0 general availability.
Anyhow, up today are just some simple benchmarks from the Intel Core i7 3770K "Ivy Bridge" system running some different Fortran/C/C++ tests while the CFLAGS/CXXFLAGS were set to "-O3 -march=native." The GCC 4.8.0 20120617 snapshot was compared to GCC 4.6.3 and GCC 4.7.1. Full details are available on OpenBenchmarking.org.
Continue exploring the results at OpenBenchmarking.org.