> This find | xargs mawk | mawk pipeline gets us down to a runtime of about 12 seconds, or about 270MB/sec, which is around 235 times faster than the Hadoop implementation.

https://adamdrake.com/command-line-tools-can-be-235x-faster-...