> On 4/8/10 14:10 , Jan Hubicka wrote: > > > So I think tying WHOPR and profile feedback too close together is a mistake. > > Sorry, I didn't mean that. My intent is to make whopr/lto use profiling > information if it is available. Much like we do with other optimization > decisions. They transparently become more precise in the presence of > profiling information.
Both LTO and WHOPR now works with profiling pretty well (and I've switched our SPEC tester to use it http://gcc.opensuse.org/SPEC/CINT/sb-frescobaldi.suse.de-fdo-64-FDO/recent.html http://gcc.opensuse.org/SPEC/CFP/sb-frescobaldi.suse.de-fdo-64-FDO/index.html it was switched in November). I see now that GCC is failing, will investigate why. The anoyance is that you really need to compile twice since instrumentation happens at compilation stage and is also read at compilation stage and just streamed to .o files. It would be better to require re-linking only. On the other hand I think if user is is keen enough to use profile feedback and LTO, he is keen to do full recompilation too, so this is not the most important missing feature of LTO we need to deal with. Honza > > > Diego.
