Gavin Thomas Nicol wrote:
> Could be... but in functional languages, there are a lot of
> optimizations that you can perform, to the point where in
> some cases, compiled functional code can be much faster than
> C/C++.
Just checking, we all here are aware of
http://www.google.de/search?q=cache:www-aig.jpl.nasa.gov/public/home/gat/lisp-study.html+lisp+performance&hl=de
(since the original
http://www-aig.jpl.nasa.gov/public/home/gat/lisp-study.html
is down) and
http://www.ai.mit.edu/docs/articles//good-news/good-news.html
right?
> For that matter, I wrote a SECDR machine a long time ago, and
> the interpreter is so tiny is *easily* fits in CPU caches, so
> it is really, really fast. I bet it'd eat most JAVA VMs for
> lunch, even with a JIT.
>
> That kinds of reminds me of FORTH... but that's probably
> another thread ;-)
A modern MISC Forth CPU will easily fit into an FPGA.
In fact it does, I've seen VHDL versions of functional
CPUs making the rounds on a few mailing lists.
In case any of you have not been tracking what Chuck has
been doing in the last half decade, have a gander at:
http://www.ultratechnology.com/dindex.htm
Forth *might* come back, through the following pathways:
* embedded RAM CPUs cum networking (if all you got is a few MBit
RAM, you'd want to use a CPU which won't take more than 20 k
transistors, and which runs very tight (threaded) code)
* FPGAs and printable hardware (i.e. where you print functional
circuits using an inkjet with polymer and semiconductor inks)
* nanotechnology (it would be nice to design a MISC CPU
in a cube full of bucky transistors)
This archive was generated by hypermail 2b29 : Sun Apr 29 2001 - 20:26:19 PDT