Benchmark: timing 100 iterations of readbig, whilelp...
   readbig: 25 wallclock secs (25.21 usr +  0.00 sys = 25.21 CPU) @  3.97/s (n=100)
   whilelp: 157 wallclock secs (156.71 usr +  0.00 sys = 156.71 CPU) @  0.64/s (n=100)