"Diehl, Jeffrey" <jdiehl@sandia.gov> writes:
> Ok, can someone explain to me why this first query might run faster than the
> second?
> select src,dst,count(dst) from data;
> select src,dst,count(*) from data;
Hmm, I'd expect the second to be marginally faster. count(*) counts the
number of rows matching the WHERE condition; count(foo) counts the
number of rows matching the WHERE condition for which foo is not NULL.
So count(foo) ought to involve a couple extra cycles to test for
non-NULL-ness of the specified field. But it's hard to believe you
could measure the difference --- what results are you getting?
regards, tom lane