norseman wrote:
<div class="moz-text-flowed" style="font-family: -moz-fixed">Marco
Mariani wrote:
Gediminas Kregzde wrote:
def doit(i):
pass
def main():
a = [0] * 10000000
t = time()
map(doit, a)
print "map time: " + str(time() - t)
Here you are calling a function ten million times, build a list with
of ten million None results, then throw it away.
def main2():
t = time()
a = [0] * 10000000
for i in a:
pass
print "loop time: " + str(time() - t)
Here you do nothing but iterating 'i' over the 'a' list.
main() # takes approximately 5x times longer than main2()
main2()
I'm wondering were is catch?
Function calls are not free in python. They cost a lot more than they
do in C, Delphi or other languages.
Gediminas - You did a good job of testing the overhead. All good
programmers do that!
Steve
If the task is to test the overhead, then the two loops need to be
equivalent. The OP's conclusion, confirmed by many of you, is that
map() is lots slower than for. But if you add in the call overhead in
the for loop, they come out within 2 percent. If you just want to
compare map() time to for time, use functions:
from time import time
def doit(i):
pass
def main():
a = [0] * 10000000
t = time()
map(doit, a)
print "map time: " + str(time() - t)
def main2():
a = [0] * 10000000
t = time()
for i in a:
doit(i)
print "loop time: " + str(time() - t)
main()
main2()
gives:
map time: 4.875
loop time: 4.79600000381
--
http://mail.python.org/mailman/listinfo/python-list