It appears that MiniSQLBackup (MSB) is spending a lot more time on the compression phase compared to writing the backups out to disk.  The I/O counters are extremely low in this case.

As Wes suggested, it is due to the compression method used.  MSB offers 2 distinct compression methods, where one is typically slower but compresses better compared to the other.  However, there are exceptions where the reverse holds true.

This may have been one such exception, where using the 'slower' method may have been faster.  The 'forced' randomness of the data may have been a factor here.

If you wait for part 2 of the shootout, you would see more comparable results.  This is because the database used is a CRM database containing real-world data.