On 1 March 2010 21:56, Max Bolingbroke <batterseapo...@hotmail.com> wrote: > I forgot to mention: I wonder if it might be worth looking at link > time optimisation, ala llvm-gcc? i.e. store LLVM bitcode in the .o > files, and substitute the llvm linker for standard ld in order to > generate assembly at link time. I suspect that GHC itself is getting > most of the cross-module inlining opportunities, but checking might be > a relatively cheap experiment to do.
Actually we both forgot :). This is on my list of todo's for the LLVM back-end (I just checked then and the rest are more mundane, cleaning tasks). This is a pretty interesting and doable project I think, you may want to look at this doc: http://www.llvm.org/docs/GoldPlugin.html This is also a project which scales in difficulty quite well. It would be fairly easy to get part of the code using LTO but quite difficult to get the entire ghc code base using LTO since llvm (and so llvm-gcc and clang) don't support all the gcc extensions the the RTS uses. _______________________________________________ Cvs-ghc mailing list Cvs-ghc@haskell.org http://www.haskell.org/mailman/listinfo/cvs-ghc
