To be able to solve 8x8 Knight's tour puzzle in a feasible amount of time Warnsdorff's rule is probably a must.
I've created a program in B-Prolog which solves the puzzle quite fast. If you need the program to be in some other Prolog - it's not too hard to translate it or just use some ideas from it.
knight_moves(X, Y, NewX, NewY) :-
( NewX is X - 1, NewY is Y - 2
; NewX is X - 1, NewY is Y + 2
; NewX is X + 1, NewY is Y - 2
; NewX is X + 1, NewY is Y + 2
; NewX is X - 2, NewY is Y - 1
; NewX is X - 2, NewY is Y + 1
; NewX is X + 2, NewY is Y - 1
; NewX is X + 2, NewY is Y + 1 ).
possible_knight_moves(R, C, X, Y, Visits, NewX, NewY) :-
knight_moves(X, Y, NewX, NewY),
NewX > 0, NewX =< R,
NewY > 0, NewY =< C,
+ (NewX, NewY) in Visits.
possible_moves_count(R, C, X, Y, Visits, Count) :-
findall(_, possible_knight_moves(R, C, X, Y, Visits, _NewX, _NewY), Moves),
length(Moves, Count).
:- table warnsdorff(+,+,+,+,+,-,-,min).
warnsdorff(R, C, X, Y, Visits, NewX, NewY, Score) :-
possible_knight_moves(R, C, X, Y, Visits, NewX, NewY),
possible_moves_count(R, C, NewX, NewY, [(NewX, NewY) | Visits], Score).
knight(R, C, X, Y, Visits, Path) :-
length(Visits, L),
L =:= R * C - 1,
NewVisits = [(X, Y) | Visits],
reverse(NewVisits, Path).
knight(R, C, X, Y, Visits, Path) :-
length(Visits, L),
L < R * C - 1,
warnsdorff(R, C, X, Y, Visits, NewX, NewY, _Score),
NewVisits = [(X, Y) | Visits],
knight(R, C, NewX, NewY, NewVisits, Path).
| ?- time(knight(8, 8, 1, 1, [], Path)).
CPU time 0.0 seconds.
Path = [(1,1),(2,3),(1,5),(2,7),(4,8),(6,7),(8,8),(7,6),(6,8),(8,7),(7,5),(8,3),(7,1),(5,2),(3,1),(1,2),(2,4),(1,6),(2,8),(3,6),(1,7),(3,8),(5,7),(7,8),(8,6),(7,4),(8,2),(6,1),(7,3),(8,1),(6,2),(4,1),(2,2),(1,4),(2,6),(1,8),(3,7),(5,8),(7,7),(8,5),(6,6),(4,7),(3,5),(5,6),(6,4),(4,3),(5,5),(6,3),(5,1),(7,2),(8,4),(6,5),(4,4),(3,2),(5,3),(4,5),(3,3),(2,1),(1,3),(2,5),(4,6),(3,4),(4,2),(5,4)]
yes