On Thu, 28 May 2015 06:03 am, Mark Lawrence wrote: > You make the statement regarding Python3 "even while it is 10 to 20 > percent slower". Where is your evidence to support this statement?
Its well known that Python 3 is generally slower than Python 2. Cecil's claim shouldn't be controversial, any more than "Python is generally about 10 to 100 times slower than C" should be controversial. But for what it's worth: [steve@ando ~]$ python2.7 -m timeit "i=1; L=sorted([30,20,50,10,40]); L[i+1]" 1000000 loops, best of 3: 1.54 usec per loop [steve@ando ~]$ python3.3 -m timeit "i=1; L=sorted([30,20,50,10,40]); L[i+1]" 1000000 loops, best of 3: 1.62 usec per loop which is an 8% slowdown. On the other hand: [steve@ando ~]$ python2.7 -m timeit "i='1'; L=sorted([30,20,50,10,40]); L[int(i)+1]" 100000 loops, best of 3: 2.6 usec per loop [steve@ando ~]$ python3.3 -m timeit "i='1'; L=sorted([30,20,50,10,40]); L[int(i)+1]" 100000 loops, best of 3: 2.29 usec per loop which is a 12% speed up. So the lesson is, micro-benchmarks are not a good guide to whole-application benchmarks. -- Steven -- https://mail.python.org/mailman/listinfo/python-list
