Thanks, tybalt89. Yes, it runs faster :) completing in less than 0.2 seconds. I updated the MCE demonstration.

# captured UNIX time
C++ 1.0          : 0.450s
C++ fast_io      : 0.291s
Perl MCE 64 thds : 0.252s

# after tweaks: CPU affinity and slurp IO
Perl MCE 64 thds : 0.218s

# more tweaks: using an ARRAY and unpack
Perl MCE 64 thds : 0.192s

$ time perl rtoa-pgatram-mce.pl t1.txt >mce.tmp
rtoa pgatram start
time 0.165 secs

real    0m0.192s
user    0m7.596s
sys     0m0.420s
[download]