Tue, 02 Mar 2010 15:33:58 -0500, bearophile wrote: > retard: >> Can you cite any sources? I'd like to know the cases where GCC is >> faster than the don's new approach. > > The thesis, page 52 (page 61 of the PDF): > http://www.cse.unsw.edu.au/~pls/thesis/davidt-thesis.pdf
Did you notice that he kept improving the backend interoperability after the thesis was done? The full potential hasn't been realized yet. Also the thesis doesn't mention Don's genetic algorithm stuff or the stream fusion optimization.
