Once upon a time, a profile of awk revealed that this loop:

for (i = 0; i < MAXFLD; i++)
    clearfield(i);

was taking 30% of the total execution time.

To allow for large inputs, MAXFLD was set to 200, but for most data, the actual number of fields is more like 2-3.

Modified the code to keep track of used fields:

for (i = 0; i < numfields; i++)
    clearfield(i);

Caused awk to run 25% faster   (along with the 1000's of programs using awk)