Benchmark: timing 1000 iterations of Way1A, Way2A, Way2B, Way3A, Way4A, Way4B, Way5A, Way5B, Way6A...
     Way1A: 32 wallclock secs ( 0.93 usr +  0.18 sys =  1.11 CPU) @ 899.28/s (n=1000)
     Way2A: 12 wallclock secs ( 1.27 usr +  0.19 sys =  1.46 CPU) @ 684.46/s (n=1000)
     Way2B: 13 wallclock secs ( 1.18 usr +  0.12 sys =  1.30 CPU) @ 768.05/s (n=1000)
     Way3A: 26 wallclock secs ( 1.00 usr +  0.23 sys =  1.23 CPU) @ 811.03/s (n=1000)
     Way4A: 16 wallclock secs ( 1.36 usr +  0.16 sys =  1.52 CPU) @ 657.03/s (n=1000)
     Way4B: 13 wallclock secs ( 1.30 usr +  0.14 sys =  1.44 CPU) @ 693.48/s (n=1000)
     Way5A: 13 wallclock secs ( 0.92 usr +  0.14 sys =  1.06 CPU) @ 942.51/s (n=1000)
     Way5B:  7 wallclock secs ( 0.90 usr +  0.17 sys =  1.07 CPU) @ 932.84/s (n=1000)
     Way6A:  8 wallclock secs ( 1.09 usr +  0.23 sys =  1.32 CPU) @ 756.43/s (n=1000)
       Rate Way4A Way2A Way4B Way6A Way2B Way3A Way1A Way5B Way5A
Way4A 657/s    --   -4%   -5%  -13%  -14%  -19%  -27%  -30%  -30%
Way2A 684/s    4%    --   -1%  -10%  -11%  -16%  -24%  -27%  -27%
Way4B 693/s    6%    1%    --   -8%  -10%  -14%  -23%  -26%  -26%
Way6A 756/s   15%   11%    9%    --   -2%   -7%  -16%  -19%  -20%
Way2B 768/s   17%   12%   11%    2%    --   -5%  -15%  -18%  -19%
Way3A 811/s   23%   18%   17%    7%    6%    --  -10%  -13%  -14%
Way1A 899/s   37%   31%   30%   19%   17%   11%    --   -4%   -5%
Way5B 933/s   42%   36%   35%   23%   21%   15%    4%    --   -1%
Way5A 943/s   43%   38%   36%   25%   23%   16%    5%    1%    --