🚀 go-pugleaf

In article <2217@megaron.arizona.edu> debray@arizona.edu (Saumya Debray) writes:
>Talking about implementations, I was quite surprised to see, at SLP 87
>earlier this month, at least two high-powered commercial Prolog systems
>suffer a factor of 3 slowdown when I changed append/3 from
>
>	append([],L,L).
>	append([H|L1], L2, [H|L3]) :- append(L1, L2, L3).
>to
>	append([],L,L).
>	append([H|L1], L2, L) :- L = [H|L3], append(L1, L2, L3).
>
>Their performance, on naive reverse on a Sun-3/75, dropped from about
>140 KLIPS for the "usual" append to about 40 KLIPS for my "mutant"
>append.  You think people would do a better job of handling 
>inline predicates and temporaries!

In at least one "high-powered commercial Prolog system" naive reverse
suffers a 35% slowdown (on a SUN) when you change append/3 to

	append([],L,L).
	append([H|L1], [H|L2], L3) :- append(L1, L2, L3).

i.e. change the order of the second and third arguments! The slowdown of append
itself is obviously greater, although I haven't made any measurements.
Apparently poor handling of inline predicates isn't the only thing that's
going on...

Lars-Henrik Eriksson					lhe@sics.se
Swedish Institute for Computer Science
Box 1263
S-163 13  SPANGA, SWEDEN