If one has a table with two columns like:

MyTable:

F_ID	Name
1		ABC
1		ABC
1		DEF
2		ABC
2		ABC
2		DEF
3		DEF
3		ABC
3		DEF

Where instead of 3 names with the same foreign ID there could be hundreds.

If One wishes to search for "ABC" returning only one match per F_ID, what is the most efficient way of doing it?

One could do:

SELECT F_ID,Name FROM MyTable WHERE Name = 'ABC' GROUP BY F_ID

However, this will create a temporary table with 

1		ABC
1		ABC
2		ABC
2		ABC
3		ABC

And then remove duplicates to get 

1		ABC
2		ABC
3		ABC

Which doesn't scale well in a scenaro where thare are a million records with f_id = 1

Alternatively one could build up a temporary table using LIMIT 1 queries, and selecting a new row where F_ID is not in the accumulated table. This performs infinitely better than GROUP BY in the worst case scenarios, but the average case is very bad.

Any ideas?