When all columns in the WHERE clause are fully indexed then the fastest method is to run 2 selects:
SELECT * FROM table WHERE col=x LIMIT 20
SELECT COUNT(*) FROM table WHERE col=x
This allows the Count to be found from just the index
SQL_CALC_FOUND_ROWS on the other hand uses filesort/temporary tables and not the index and consequently is slower
This is a known bug/feature request http://bugs.mysql.com/bug.php?id=18454
On the other hand, experience/benchmarking has shown me that the situation can be reversed when columns in the WHERE clause are not fully indexed and COUNT(*) is forced into a table scan.
By the way, to benchmark your queries you use EXPLAIN