Honza,

> I will also look into the estimate_size ICE reported today.

This has been fixed by revision 172603 and now the thresholds are
the same with/without -flto. For the fine tuning I have posted
some results on pr48636.

Cheers,

Dominique

Reply via email to