Re: LW compiler optimizations

I don't do any DSP programming,  but it seems like a standard interface 
for MAC could be useful Apparently C has a standard interface called 
"fma()". Most computers have hardware for this. In addition, some higher 
level interfaces could also be standardized, where the operating system 
or other libraries are (FFI or otherwise) called. Those could be 
parameterized so the user can decide which implementation they want, 
with a fallback to Lisp code.

In the worst case (no actual compiler hardware support), this would give 
semantics to the code, so that when support is maybe available later, 
it's easier to see where it could be called. There might be support on 
some platforms and not others, but the code would be portable. Even if 
it had to name platform/vendor-specific features in conditionals, the 
high-level semantics improve the quality of the code in terms of 
documentation and portability.

This could be developed in the community and shared via Quicklisp. 
That's how we generally "extend" Common Lisp.

As for assembly language in Lisp, that's something that has historically 
been available in some implementations, one way or another.  But that's 
what #+ conditionals are for, and you then hide those inside the 
abstractions in the portability libraries.

The DSP Lisp hackers here are sophisticated enough to know all this. I'm 
not one of them. I'm just jumping in to say it out loud. We have every 
opportunity to do this better than the other languages, especially with 
regards to static/dynamic algorithm/implementation selection.



_______________________________________________
Lisp Hug - the mailing list for LispWorks users
[email protected]
http://www.lispworks.com/support/lisp-hug.html