Michael Kreim wrote: > Peter Otten wrote: >> Move it into a function; this turns a and i into local variables. >> >> def f(): >> imax = 1000000000 >> a = 0 >> for i in xrange(imax): >> a = a + 10 >> print a >> f() > > Wow. It is still slower than Matlab, but your suggestion speeds up the > code by ca 50%. > But I do not understand why the change of a global to a local variable > gives such a big difference.
Basically the local namespace is a C array where accessing an item is just pointer arithmetic while the global namespace is a Python dictionary. There may be optimisations for the latter. If you can read C have a look at the LOAD/STORE_FAST and LOAD/STORE_GLOBAL implementations for the gory details: http://svn.python.org/view/python/trunk/Python/ceval.c?view=markup Peter -- http://mail.python.org/mailman/listinfo/python-list
