Yeah my CHR solution doesn’t scale so well:
?- set_prolog_flag(stack_limit, 2_147_483_648).
true.
?- time(count(64,X)), write(X), nl, fail.
% 901,974,873 inferences, 84.953 CPU in 85.473 seconds (99% CPU, 10617324 Lips)
Compare to the mode directed tabling:
But I see the same accepted answer for both. So both seem to be correct.