Hmmm I wasn't expecting a huge speedup but was hoping for a few percentage points at least. Can you fork GHC on github and put your patch up on there and then send me a link please? This is just a far easier workflow for me to review and merge patches than email.
Cheers, David On 13 October 2011 01:37, Erik de Castro Lopo <[email protected]> wrote: > David Terei wrote: > >> Yes that will work and be a large scale test. Easier though is to just >> use the nofib benchmark 'included' with ghc. >> >> Thanks for having a crack at this Erik. > > Ok, after being distracted by some work related stiff I'm back > on this issue. > > I used the compile time part of the nofib benchmark and compared > three configurations: > > fasm-1 - Compile with -fasm before my llvm ppr changes. > fllvm-1 - Compile with -fllvm before my llvm ppr changes. > fllvm-ppr-1 - Compile with -fllvm after my llvm ppr changes. > > The nofib-analyse output comparing all three and just the two > -fllvm versions are included below which shows that my ppr > shows a slight compile speed up but going via the llvm backend > is still significantly slower incomparison to via the NCG. > > I suspect that better benchmarking could be acheived by testing the > LLVM output in isolation. > > I have also attached the patch that I tested. > > Cheers, > Erik > -- > ---------------------------------------------------------------------- > Erik de Castro Lopo > http://www.mega-nerd.com/ > > _______________________________________________ > Cvs-ghc mailing list > [email protected] > http://www.haskell.org/mailman/listinfo/cvs-ghc > > _______________________________________________ Cvs-ghc mailing list [email protected] http://www.haskell.org/mailman/listinfo/cvs-ghc
