Thank you Dan. I've always been impressed with Gambit, but I guess it hadn't really penetrated my skull that Marc was using C as his target. I'll chek it out again, and see if I can find his Reno talk. I saw some good slides on his webpages before.<br>
<br>From what I've heard about GHC, some later news (May 2010) announced that David Terei actually did implement an LLVM backend as a honors thesis. As a part of it, Chris Lattner and the LLVM group agreed to implement a specific optimization in LLVM that the Haskell folks needed to make their code fast. As a result I think the LLVM backend is now 10-30% faster than GHC. Apparently it ships standard as an option as of ghc 7.0.<br>
<br>Don Stewart talks about it<br><a href="http://donsbot.wordpress.com/2010/02/21/smoking-fast-haskell-code-using-ghcs-new-llvm-codegen/">http://donsbot.wordpress.com/2010/02/21/smoking-fast-haskell-code-using-ghcs-new-llvm-codegen/</a><br>
<br>David did a writeup on it here<br><a href="http://blog.llvm.org/2010/05/glasgow-haskell-compiler-and-llvm.html">http://blog.llvm.org/2010/05/glasgow-haskell-compiler-and-llvm.html</a><br><br>full details<br><a href="http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/Backends/LLVM/Design">http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/Backends/LLVM/Design</a><br>
<br><br>