[Kurt Johannes]

Hi,

I am new to Python so I implemented some simple applications to see the performance of Python in comparison to Java and Symbian. Therefore I just programmed some for loops with some simple mathematical operations (subtraction, multiplication) to see the difference to other platforms.

I was suprised that Python is really slow compared to other languages. Of course, I know, it is interpreted but it is ~15 times slower than java and much more slower than Symbian.

Is it really that slow or am I doing something wrong?

Code snippet:

v1 = [..20 float values]
v2 = [..20 float values]

for i in range(100000):
for j in range(20):
d = v1[j]-v2[j]
dd = d*d

[jplauril]

Try using xrange instead of range. That may improve performance a little bit. But yes, Python is slower than Java or C++. You should always try to perform mathematical heavy lifting using native modules instead of coding it elementwise in Python.

[miohtama]

Originally Posted by Kurt Johannes

Hi,

I was suprised that Python is really slow compared to other languages. Of course, I know, it is interpreted but it is ~15 times slower than java and much more slower than Symbian.

Instead of for loops try use map, filter and other built-in Python mapping operations. They lift the loop iteration to the native code and perform faster.

If your math code is performance sensitive, move crucial loops to the native C++ extension. This way you get the absolute processor speed, the same as in native C.

Cheers,
Mikko