This is not an issue so much as my thoughts as an applied mathematician, and Intel Device Driver tester for several years out of Folsom, CA.
You may find that it would be more descriptive to provide % std Dev instead of the actual Std Dev.
So for the Table benchmarks/bench-x-small.log

% Std Dev = 100* std Dev/Arithmetic Mean Average is what I believe you mean by Average

Type	Average ops/sec	% Std. Dev. ops/sec
Set	417,070	1.9
Delete	583,134	1.25
Get	7,891,966	2.66

Now you can immediately see that the std dev for Get is TWICE that of Delete and Set and perhaps lead you to ask the question why ?? What is going on with Get that the Std Dev is twice as wide for Get than for Set and Delete

For your 10,000,000 iteration test
benchmarks/bench-small.log

Type	Average ops/sec	% Std. Dev. ops/sec
Set	360,777	0.86
Delete	466,481	0.89
Get	6,171,403	2.82

Notice the decrease in the % Std Dev for Set and Delete WHILE THE Get % std dev DID NOT DECREASE at all
This is an immediate red flag that should be investigated

This is your reputation that is on the line, so I suggest you investigate the 2 tables benchmarks/bench-large.log as well I think they follow a similar pattern.
My best wishes to you and your loved ones