Re: [Swig-user] swig speed is not as fast as the original C++???

[Pujo A. <aj...@gm...>]

Thanks Marcelo,

You're right, because looping in python is frustrating. Now it is fast.
I have a few problems:
my compiler doesn't allow me to define type of variable like this:
    for (int i  =3D 0; i<10; i++){
       //mycode
    }

I should use :
    int i =3D 0;
    for (i  =3D 0; i<10; i++){
       //mycode
    }

Is that standard?



Thanks,
pujo



On 12/5/05, Marcelo Matus <mm...@ac...> wrote:
>
> You are not just comparing the swig/C++, you have a huge for loop in
> there.
>
>
> Try the other way, create a function in C++ with the loop inside
> and the call it from python.
>
> also,look at the profiletest_runme.py in the
> SWIG/Examples/test-suite/python for
> ideas of comparing using loops.
>
> Marcelo
>
>
> Pujo Aji wrote:
>
> > Hello All, Marcelo,
> >
> >
> > I manage to use visual studio express C++ (.Net Framework 2) compiler
> > with swig and python.
> > By tweaking pyConfig, installing SDK Platform etc.
> > I can run the runme.py in "simple" example.
> >
> > I have  2 questions :
> > 1. Concerning compiling simple.c and simple_wrap.c:
> >     The last thing that I need is simple.py and _simple.dll is this
> > correct ?
> >     Because in the tutorial it mentioned simple.o and simple.so which
> > I can't find these files in my computer (win xp)
> >
> > 2. Concerning swig speed.
> > I found something very interesting.
> > I create a function in C++ and use swig to wrap it.
> > After getting simple.dll and simple.py. I can use it in my main python
> > program (runme.py).
> > I compare the speed between : pure C++, C++wrap with swig,  pure
> > python, and python with psyco.
> > This is the result:
> >
> > *pure C++ : 1 second
> >
> > python without psyco:
> > C++wrap with swig: 13.76 second
> > pure python: 37.19 second
> >
> > python with psyco:
> > C++ wrap with swig: 14.55 second
> > pure python: 15 second
> >
> > Strange, I thought at the first time:
> > 1. with psyco the speed of C++ wrap with swig is almost the same speed.
> > 2. pure C++ is much much faster than C++ wrap with swig and used in
> > python ???
> >
> > Note :
> > * for pure C++, I can't test until the unit in millisecond, but I
> > minimize this problem by using problem that make pure python and
> > C++wrap with swig produce  a couple of seconds.
> >
> > Looking forward to your responds
> > Sincerely Yours,
> > pujo
> > .
> >
> >------------------------------------------------------------------------
> >
> ># file: example.py
> >
> >import example
> >import time
> >import psyco
> >psyco.full()
> >
> >def fact(n):
> >    f =3D 1
> >    while( n>1):
> >        f *=3D n
> >        n -=3D 1
> >    return f
> >
> >## Call our gcd() function
> >#
> >#x =3D 42
> >#y =3D 105
> >#g =3D example.gcd(x,y)
> >#print "The gcd of %d and %d is %d" % (x,y,g)
> >#
> >## Manipulate the Foo global variable
> >#
> >## Output its current value
> >#print "Foo =3D ", example.cvar.Foo
> >#
> >## Change its value
> >#example.cvar.Foo =3D 3.1415926
> >#
> >## See if the change took effect
> >#print "Foo =3D ", example.cvar.Foo
> >
> >start =3D time.time()
> >sum =3D 0
> >for i in range(10000000):
> >       sum+=3Dexample.fact(10)
> >stop =3D time.time()
> >print sum
> >print 'wrap c++ time,', stop-start
> >
> >
> >start =3D time.time()
> >sum =3D 0
> >for i in range(10000000):
> >       sum+=3Dfact(10)
> >stop =3D time.time()
> >print 'python time,', stop-start
> >
> >
> >
> >
> >
> >
> >
> >
>
>