search for: record_edge

...separate pass later. > I'm not sure why you'd add it as a separate pass, it is about 80-150 lines of code, and adding it as as a separate pass requires you to do things like induction variable detection + etc all over again. See http://gcc.gnu.org/ml/gcc-patches/2007-01/msg01035.html, record_edge, apply_lftr_edge, follow_lftr_edge and perform_lftr The LLVM development documentation suggests that new work be committed > piecemeal over time. > > Sure, but that doesn't mean you should commit something that is going > LLVM does have an optimization pass, -instcombine, which w...

On 11/13/10 11:05 PM, Daniel Berlin wrote: > > A big downside of the current LSR algorithm is it's slow. I had > initially > hoped that some of the heuristics would protect it better, but the > problem is > more complex than I had expected. I haven't done any measurements, > but it's likely that OSR is faster, which may interest some

> > A big downside of the current LSR algorithm is it's slow. I had initially > hoped that some of the heuristics would protect it better, but the problem > is > more complex than I had expected. I haven't done any measurements, > but it's likely that OSR is faster, which may interest some people > regardless > of how the output compares. > A few years