21 Oct
2006
21 Oct
'06
9:26 a.m.
Jay West wrote:
In theory, you are right. In practice I believe
you're making too much
of a generalization. In school they typically teach that interpreters
are always slower in order to explain and drive home the differences. In
the real world however, that's quite simply not always the case. I've
written both professionally, and I've seen instances that would
obviously suprise you. Now if you're comparing interpreted languages vs.
written by hand assembler, I could agree with you. But when a compiler
is the one generating the object, well, you may be suprised at how
closely a interpreted stack machine can get to the ratio of required
hardware instructions given the platform.
Also at the time most floating point was done in software
for the smaller machines. The ratio on integer math 16 bit
to floating point may be the factor in most interpreted lanquges
being the about the same speed as compiled ones.
BASIC/09 for the 6809 was nice in that you could have both versons
compiled or interpreted from the same source. PASCAL was nice too
in that you had virtual memory ( crummy with a floppy ) so you could
run larger programs than 64k.
Jay
.