A regular index, but whether it would be of any use depends on the distribution of the data and the "clustering factor" of the index. The optimizer is quite likely to prefer a full scan over this index in many/most cases.
Distribution: Suppose your data looks like this:
select col, count(*) from mytable group by col;
Here the optimizer might choose to use the index for a query "where col='D'" but not for "where col='A'".
Clustering: If in the example above, all the 'A' rows are physically close together in the table, and ditto for other values, then the optimizer may use the index because it expects to only have to visit a small proportion of the blocks in the table. But if they are "all over the place", then it may have to visit most of the blocks and so may as well do a full scan.