BTW, I moved the articles to a different host last year,

http://donsbot.wordpress.com/2008/05/06/write-haskell-as-fas...

and the follow-on using fusion,

http://donsbot.wordpress.com/2008/06/04/haskell-as-fast-as-c...

Way cool, thanks :]

I am happy to see UNSW powering on in the FP world. Waaay back when I went there, they tried, valiantly, to teach us grubby Engineer hacker types the virtues of purity (via miranda), of I tried to ignore it and wanted to get to the good stuff that let me flip bits on a serial port. But now I am grateful for someone at least trying to "show me the way" so early on (and over time, I think it paid off).

Ah, was posted previously: http://news.ycombinator.com/item?id=190963

Still, an awesome analysis that traverses the code hierarchy from assembly to Haskell.

I personally only care about optimizations that enable someone who's a domain expert in something else to write code that performs well. If I know Haskell at an intermediate level, and I'm focusing on the equations for protein folding, or game mechanics, or rocket trajectories, will I still be able to write code that performs well? Or do I need to be a Haskell expert first?

Wasn't that what the author was getting at? The DSL for including rewrite/optimization rules in the compiler is actually accessible to the "mere mortal".

Yes and no. It's great that it's available. I'm just not seeing that it implies a workable routine for Joe Programmer.

Write code. Run. Runs out of memory after 3 hours of protein folding. Check the GHC processed code. Discover that you need to rewrite 5 functions. The premise is that your mental model wasn't good enough to predict this without the GHC processing, so rewrite, run through GHC, stare at code, rewrite again. If that whole cycle is necessary, then this does not sound workable. But I'd be thrilled to learn that I'm missing something.

To be fair, no language lets you just "write code" and have it run anywhere near optimally. The programmer always has to be aware of what the compiler is going to do and work with it.

I've written a fair amount of Haskell, and done a lot of this type of optimization, looking at the core (intermediate representation) and fiddling with pragmas, unboxing, strictness, etc. I've put Haskell on the back burner and gone back to C++ because I spent SO MUCH time doing this. Maybe I'm not being fair, and in another 5 years I'd be able to predict what GHC is going to do as well as I can predict what a C++ compiler will do. But for now, I can't justify the expenditure in time it takes to get a large piece of Haskell code (more than a microbenchmark) to perform at the C/C++ level. If you write idiomatic Haskell you're tracking down stack overflows (the Haskell kind, not the C kind) and misfired optimizations, and if you're writing C-in-Haskell you're better off just writing C. In the end, I think Haskell just isn't (yet) the best language for what I was doing (image processing, graphics, 3d modeling, etc.)

(The short reply to your comment is: yes, you the application writer have to be very aware of what GHC is doing if you want C-level performance.)

Thank you. That was the impression I got. I'm still curious to hear if anyone else had a different experience.

I'm not so much looking for what it takes to get code running optimally, as much as having idiomatic code run with "reasonable" performance.