Benchmark results from testing several python hash libs against str.__hash(), on my MBP Retina, 15-inch, Early 2013. This was run under Python 2.7. using various python hash functions. MurmurHash3 was used for the Murmur tests.

python-hash-bench

Note: "memo" is just testing constructing a string, to prove 
that strings are not hashed at create time, but memoized with 
__hash__(). The functions construct new strings each time, 
with some variant of:

    '24058098:33d2a96e-ffad-11e2-b14b-8acd18156acf' * m

Where m is passed into the test function.

---------------------------------------------

In [151]: %timeit test_memo()
1000000 loops, best of 3: 280 ns per loop

In [152]: %timeit test_builtin()
1000000 loops, best of 3: 276 ns per loop

In [153]: %timeit test_murmur()
1000000 loops, best of 3: 386 ns per loop

In [154]: %timeit test_xx()
1000000 loops, best of 3: 358 ns per loop

In [155]: %timeit test_memo(2)
1000000 loops, best of 3: 325 ns per loop

In [156]: %timeit test_builtin(2)
1000000 loops, best of 3: 414 ns per loop

In [157]: %timeit test_murmur(2)
1000000 loops, best of 3: 441 ns per loop

In [158]: %timeit test_xx(2)
1000000 loops, best of 3: 409 ns per loop

In [159]: %timeit test_memo(5)
1000000 loops, best of 3: 442 ns per loop

In [160]: %timeit test_builtin(5)
1000000 loops, best of 3: 700 ns per loop

In [161]: %timeit test_murmur(5)
1000000 loops, best of 3: 655 ns per loop

In [162]: %timeit test_xx(5)
1000000 loops, best of 3: 571 ns per loop