>On Wed, 2006-09-13 at 08:34 +0200, mathieu lacage wrote: >> hi, >> Maybe it is the idea of writing an IPA pass operating on SSA which is >> just plain braindead in which case it would be nice for someone to tell >> me so :)
>It is not braindead except GCC currently does not support that on the >mainline. On the ipa-branch you can do it, in fact inlining is already >done on SSA there and works. > >Thanks, >Andrew Pinski Right. And also Interprocedural constant propagation and matrix flattening are ssa IPA optimizations implemented on ipa branch. Razya
