Throwing my hat in the ring:
1) First we execute the NOT function on a set of integers,
evaluating non-zeros to 0 and zeros to 1
2) Then we search for the MAX in our original set of integers
3) Then we multiply each number in the set generated in step 1 by the MAX found in step 2, setting ones as 0 and zeros as MAX
4) Then we add the set generated in step 3 to our original set
5) Lastly we look for the MIN in the set generated in step 4
{=MIN((NOT(A1:A5000)* MAX(A1:A5000))+ A1:A5000)}
If you know the rough range of numbers, you can replace the MAX(RANGE) with a constant. This speeds things up slightly, still not enough to compete with the faster functions.
Also did a quick test run on data set of 5000 integers with formula being executed 5000 times.
{=SMALL(A1:A5000,COUNTIF(A1:A5000,0)+1)}
1.700859 Seconds Elapsed |
5,301,902 Ticks Elapsed
{=SMALL(A1:A5000,INDEX(FREQUENCY(A1:A5000,0),1)+1)}
1.935807 Seconds Elapsed |
6,034,279 Ticks Elapsed
{=MIN((NOT(A1:A5000)* MAX(A1:A5000))+ A1:A5000)}
3.127774 Seconds Elapsed |
9,749,865 Ticks Elapsed
{=MIN(If(A1:A5000>0,A1:A5000))}
3.287850 Seconds Elapsed |
10,248,852 Ticks Elapsed
{"=MIN(((A1:A5000=0)* MAX(A1:A5000))+ A1:A5000)"}
3.328824 Seconds Elapsed |
10,376,576 Ticks Elapsed
{=MIN(IF(A1:A5000=0,MAX(A1:A5000),A1:A5000))}
3.394730 Seconds Elapsed |
10,582,017 Ticks Elapsed