Serhiy Storchaka added the comment:
Here is even faster patch. When there are no dummies in source set we can just
dump a table, placing entries at the same indices.
$ ./python -m timeit -s "s = set(range(10**4))" -- "frozenset(s)"
Unpatched: 1000 loops, best of 3: 658 usec per loop
Patched: 1000 loops, best of 3: 631 usec per loop
$ ./python -m timeit -s "s = {i+(j<<64) for i in range(10**3) for j in
range(10)}" -- "frozenset(s)"
Unpatched: 100 loops, best of 3: 6.72 msec per loop
Patched: 1000 loops, best of 3: 930 usec per loop
$ ./python -m timeit -s "s = {i+(j<<64) for i in range(10**2) for j in
range(10**2)}" -- "frozenset(s)"
Unpatched: 100 loops, best of 3: 14 msec per loop
Patched: 1000 loops, best of 3: 1.12 msec per loop
To test other branch we should add dummy entry: s.add(-1); s.discard(-1).
$ ./python -m timeit -s "s = {i+(j<<64) for i in range(10**3) for j in
range(10)}; s.add(-1); s.discard(-1)" -- "frozenset(s)"
Unpatched: 1000 loops, best of 3: 661 usec per loop
Patched: 1000 loops, best of 3: 643 usec per loop
$ ./python -m timeit -s "s = {i+(j<<64) for i in range(10**3) for j in
range(10)}; s.add(-1); s.discard(-1)" -- "frozenset(s)"
Unpatched: 100 loops, best of 3: 6.8 msec per loop
Patched: 100 loops, best of 3: 2.1 msec per loop
$ ./python -m timeit -s "s = {i+(j<<64) for i in range(10**2) for j in
range(10**2)}; s.add(-1); s.discard(-1)" -- "frozenset(s)"
Unpatched: 100 loops, best of 3: 14 msec per loop
Patched: 100 loops, best of 3: 2.71 msec per loop
----------
Added file: http://bugs.python.org/file37808/set_faster_copy_2.patch
_______________________________________
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/issue23290>
_______________________________________
_______________________________________________
Python-bugs-list mailing list
Unsubscribe:
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
