Mailman 3 April 2012 - scikit-image

Re: [OT] pass a C callback function to a cythom module via python
by Nadav Horesh April 2, 2012

April 2, 2012

I do not have a benchmark code yet. I am working now on unpythonising the in-loop buffer copy (the line "neigh[:] = self.padded[i:i+ks0, j:j+ks1].ravel()" in filter_iter.pyx)/ When I'll finish this and include your suggestion, I'll be ready for the benchmarking. My experience with the bilateral filter showed that yellowish lines do not comply with nogil, but it is easy to verify again. Nadav. בתאריך 1 באפריל 2012 19:47, מאת Stéfan van der Walt <stefan(a)sun.ac.za>: > Hi Nadav > > On Sun, Apr 1, 2012 at 6:42 AM, Nadav Horesh <nadavh.horesh(a)gmail.com> > wrote: > > A nice idea. As is, it can not replace the link to C code linking > because it > > contains to much yellow lines ("cython -a" output) in the main loop, so > > there is a speed penalty, and also I suspect it wouldn't be possible to > > parallelise that function call. Anyway I plan to add this option. > > Could you share your benchmark code? This should be very nearly as > fast as the C pointer call. And with an appropriate "with nogil" > statement, I think it should parallelize? > > Stéfan >

1 0

Re: [OT] pass a C callback function to a cythom module via python
by Stéfan van der Walt April 1, 2012

April 1, 2012

Hi Nadav On Sun, Apr 1, 2012 at 6:42 AM, Nadav Horesh <nadavh.horesh(a)gmail.com> wrote: > A nice idea. As is, it can not replace the link to C code linking because it > contains to much yellow lines ("cython -a" output) in the main loop, so > there is a speed penalty, and also I suspect it wouldn't be possible to > parallelise that function call. Anyway I plan to add this option. Could you share your benchmark code? This should be very nearly as fast as the C pointer call. And with an appropriate "with nogil" statement, I think it should parallelize? Stéfan

1 0

Re: [OT] pass a C callback function to a cythom module via python
by Nadav Horesh April 1, 2012

April 1, 2012

A nice idea. As is, it can not replace the link to C code linking because it contains to much yellow lines ("cython -a" output) in the main loop, so there is a speed penalty, and also I suspect it wouldn't be possible to parallelise that function call. Anyway I plan to add this option. About ctypes: I am not a ctypes fond either, neither I think that my use is elegant or very portable. The ctypes is used for one purpose: load a shared library and provide a pointer to an object within that library. It seems to be that it is inevitable that there is a simpler way to to that, but I am clueless. Nadav. בתאריך 28 במרץ 2012 00:03, מאת Stéfan van der Walt <stefan(a)sun.ac.za>: > >> Hi Nadav >> >> I just had a chat with Dag, a core Cython contributor. He mentioned >> to me another alternative, which is to encapsulate the callback >> function in a class: >> >> >> https://github.com/dagss/cython-pydata12/blob/master/examples/oointegrate/i… >> >> You can import this class from Python, and pass it along as the >> callback object. There may be some work on improving this >> functionality in the future, but for now I think that's the cleanest >> route (and helps us to avoid ctypes!). >> >> Stéfan >> >> >> On Tue, Mar 20, 2012 at 1:02 AM, Nadav Horesh <nadavh.horesh(a)gmail.com> >> wrote: >> > >> > Can you explain in more details how can I implement it? Any case I am >> > attaching an scheme that may clarify my intend. Soon I'll post a code >> > demonstrating the problem. >> > >> > Thank you, >> > Nadav, >> >

1 0