sasaki@harvard.ARPA (Marty Sasaki) (07/02/84)
Just a little bit to fan the flames.
A number of years ago a group of MIT hackers did the following experiment.
A FORTRAN to LISP translater was hacked together. Then a challenge was
issued, anyone who had a FORTRAN program that ran on a Decsystem-10/20
was asked to submit it to the hackers to see if the program ran faster in
FORTRAN or in LISP (MacLISP (gosh, I didn't know that Apple had a LISP...)).
The results were surprising. I don't remember exactly, and don't have a
printout, but every program ran faster in LISP than in FORTRAN (cpu time).
These were number crunching programs. The translater was very simple and did
no optimization on the code.
Lisp is often badmouthed because it doesn't run efficiently, but in this case
it simply isn't true. I wonder how well more modern Lisp/FORTRAN
implementations fare?
Marty Sasaki
Havard University Science Center
sasaki@harvard.{arpa,uucp}