[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Lisp is faster than Prolog? A personal plea

	Good point, Paul, but I think you're missing something.  First you
plead with us not to use micro-benchmarks, then you point out (correctly)
that the strategy that one would use to write a program in Lisp instead of
Prolog can often differ.  I would think that the implication from the latter
observation is that large programs are fundamentally incomparable, and I
think that that is probably correct.

	So if you deny us micro-benchmarks, then we can not measure the
relative performance of these languages at all (or, more precisely, the
standard implementations of these languages on the 11/780).  Hence we might
as well accept the statements "Prolog is faster than Lisp" or "Lisp is faster
than Prolog" or "Lisp is faster than assembler" as essentially meaningless
statements, since we can't quantify any of them.

	Let me sputter out making one final point.  LIPS is not all that
bad a measure.  Perhaps if we called it "cycles through the append loop" or
"function calls per second" (essentially identical statements) I think most
people would agree that this is a fair measure of the performance of any
Lisp.  After all, Lisp does nothing other than call functions and manipulate

	I'm certainly not going to take issue with the rest of your letter,
which is really more directed at Sanjai's claims than mine, and walks rather
closer to debates on programming style than any sane man should dare to go.

	I remain, sir,

					Y'r obedient servant,

						Rick McGeer.