Hi Jeffrey, I'm seeing a few performance regressions similar to http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32306 and http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33315 in a port where I'm working off the 4.3 branch. These regressions are caused by the decision to stop iterating DOM on identifying jump threading oppurtunities and splitting up VRP into a separate pass based on your patch at http://gcc.gnu.org/ml/gcc-patches/2006-02/msg00586.html and as described by Richard here http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32306#c7.
I'm going to try a quick experiment by re-running DOM if jump threading is turned on in the 4.3 branch. I know there are a number of changes I need to do and I'm looking into them. I wanted to check back if you still had any record of the small number of performance regressions that you mention in the posting of the afore mentioned patch. Thanks, Ramana
