If you leave 2nd sort, it would be quicker as itab will be there in right order.
SORT itab by f1 f2 f3. READ TABLE itab WITH KEY f1 = 'A' f2 = 'B' f3 = 'C' BINARY SEARCH. READ TABLE itab WITH KEY f1 = 'A' BINARY SEARCH.
When 2nd READ TABLE is by f2, you should leave the SORT and BINARY SEARCH as well. In best case scenario, SORT + BINARY SEARCH speed is n+log(n).
Also note that worst case of sort is n log n.