Which query is faster, select count(*) or count(pk)?

Definitely select count(1) is too much faster,
The count(pk) is equal to select count(1) where pk is not null,
In my case select count(1) takes 200ms but the select count(id) takes more than 20 seconds (the table has more than 10B records)!
Also i think select count(1) and select count(*) are same