A more accurate follow-up to my hurried post of a few hours ago.
Timings on the same machine for a more truthful speed comparison, now
in reality "just" a 150x speed-up due to Cython.
Triple-braces delimit compute cells in the notebook. Notice that the
"%cython" block automatically produces the C output and an HTML file
of the source, where you can click on individual lines to see the per-
line conversion to C, with less-efficient conversions highlighted in
yellow.
Hopefully an approach with very fast code like this for internal
computations, plus fast conversions to Sage types at the user-level
might make for a good combination. Developing with the Sage types
until you get it right, and then optimizing replacements with Cython
code is a good strategy as well.
I'm sure others with more experience will have more knowledgeable
opinions or solutions. It'll be great to see some matroid theory
available in Sage - thanks for your work on this.
Rob
{{{
def intersect_python():
t = 0
for s1 in range(0,2**9):
for s2 in range(0,2**9):
if s1 & s2:
t += 1
return t
}}}
{{{
timeit("intersect_python()", number=100)
}}}
100 loops, best of 3: 31.5 ms per loop
{{{
%cython
def intersect_cython():
cdef Py_ssize_t t, s1, s2
t = 0
for s1 in range(0,2**9):
for s2 in range(0,2**9):
if s1 & s2:
t += 1
return t
}}}
{{{
timeit("intersect_cython()", number=10000)
}}}
10000 loops, best of 3: 205 µs per loop
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
