Right. I often added debug out using printf "\r%d\t", $loopVariable unless $loopVariable % 1000, but if STDOUT is buffered, then you have to wait for the buffer to fill before you "see" the first output, and then it increments in leaps and bounds. You can see this with:

c:\test>perl -lwe" Win32::Sleep( 10 ),printf qq[\r%d\t], $_ for 1 .. 1
+e6"
[download]

Which appears to display (overlaid of course):

841
1550
2232
2915
...
[download]

Update: Once you get to 4-digit numbers, the interval is 682. With 4-digits, "\r" & "\t", that is 6 * 682 = 4092. A 4kb buffer.

Whereas if you disable the buffering

c:\test>perl -lwe"$|=1;Win32::Sleep( 10 ),printf qq[\r%d\t], $_ for 1 
+.. 1e6"
[download]

You see the number increment from 1, quickly and continuously.

I guess the buffering doesn't consider "\r" as a line ending -- probably rightly. It is useful for monitoring the progress of long running processes. Leaving the buffering enabled means that the IO has a negligable effect upon the speed of the code, and has the beneficial side effect that I can 'pause' and 'resume' the long running process by using the left and right mouse buttons alternately. Mouse equivalent of ^S.