xrange vs range

[endophage]

In almost every instance you use range, you should instead be using xrange. Test 16 is especially painful to look at as range(1000) already creates an array with the integers 0 to 999, which you then iterate over to create another copy of the array.

Test 16 should really be a comparison of the two versions you have (but using xrange instead of range), vs just "return range(1000)". A quick test on my system shows a simple "return range(1000)" to be a further 3-4x faster than your fastest Test 16 variant, and substituting xrange for range achieves a 20-30% improvement over your fastest variant.

[endophage]

from timeit import timeit
>>> def a():
...     return [i for i in range(1000)]
...
>>> def b():
...     return range(1000)
...
>>> def c():
...     return [i for i in xrange(1000)]
...
>>> timeit(a, number=100000)
2.728806972503662
>>> timeit(b, number=100000)
0.723524808883667
>>> timeit(c, number=100000)
2.2922120094299316