Loop optimizer issues

[Gerald Pfeifer]

On Mon, 28 Jul 2003, Zdenek Dvorak wrote:
> I am basically done with the work on rtl-level optimizer. [...] I have
> done some testing (it bootstraps now on x86, powerpc and ia64) and
> benchmarking (there are a few regressions to investigate, see attached
> results of SPECs).

Just a note: today I compared mainline against tree-ssa, and the code
generated by tree-ssa with -O3 is 25% slower, so merging something else
which may cause performance regressions doesn't seem like a good idea.

Gerald