[Chennaipy] Mutli- threading in cython

aishwarya selvaraj aishwaryaselvaraj1708 at gmail.com
Sun Jun 3 10:01:43 EDT 2018 

Hi,
With  npstrides[1] = 2, it worked perfectly.
Thank you.
ᐧ

On Sat, Jun 2, 2018 at 1:01 PM, Aravindhan K <aravindhank1994 at gmail.com>
wrote:

> Hi,
>
> Use the version of code that is mentioned in your NOTE 4.
>
> But why are you keeping np.dtype('<'B)  instead of  np.dtype('<'H') ? in
>
> data[i,:,:] = PyArray_NewFromDescr(<PyTypeObject *> np.ndarray,
> np.dtype('<B'), 2,dims, strides,pBuf, np.NPY_C_CONTIGUOUS,     None)
>
> and change this too.
> npstrides[1] = 2 #short is two bytes long.
>
>
> Regards,
> Aravindhan K
>
> On Fri, Jun 1, 2018 at 4:32 PM, aishwarya selvaraj <
> aishwaryaselvaraj1708 at gmail.com> wrote:
>
>> Hi,
>> I tried the changes. No was not able to solve my issue. I'm attaching the
>> file. Kindly take a look at it.
>>
>> ᐧ
>>
>> On Mon, May 28, 2018 at 10:21 PM, Aravindhan K <aravindhank1994 at gmail.com
>> > wrote:
>>
>>> Hi,
>>>
>>> See the comments starts with NOTE in the attached file.
>>> Hope that may guide you find the issue.
>>>
>>> Regards,
>>> Araindhan K
>>>
>>> On Mon, May 28, 2018 at 11:35 AM, aishwarya selvaraj <
>>> aishwaryaselvaraj1708 at gmail.com> wrote:
>>>
>>>> Hi,
>>>> Yes, I was able to make use of  PyArray_NewFromDescr in my code.
>>>> Storing into a 3D array has now become a lot faster. Thank you!But the
>>>> CCD images stored using this method looks quite different from the ones
>>>> acquired earlier.
>>>> In the older method at the end of for loop, there is incrementing using
>>>>  _puc_image += aoiStride, I'm not very sure on how to do it with new
>>>> method. Could you please take a look at it?
>>>> ᐧ
>>>>
>>>> On Mon, May 28, 2018 at 11:17 AM, aishwarya selvaraj <
>>>> aishwaryaselvaraj1708 at gmail.com> wrote:
>>>>
>>>>> Hi,
>>>>> Yes, I was able to make use of  PyArray_NewFromDescr in my code.
>>>>> Storing into a 3D array has now become a lot faster. Thank you!But
>>>>> the CCD images stored using this method looks quite different from the ones
>>>>> acquired earlier.
>>>>> In the older method at the end of for loop there is incrementing
>>>>> using  _puc_image += aoiStride, I'm not very sure on how to do
>>>>> it with new method. Could you please take a look at it? I'm not very
>>>>> sure what am I missing here.
>>>>> ᐧ
>>>>>
>>>>>>
>>>>>>> dims[0] = self.aoiWidth
>>>>>>
>>>>>> dims[1] = self.aoiHeight
>>>>>>
>>>>>>                         #method 1
>>>>>>
>>>>>> data[i,:,:] = PyArray_NewFromDescr(<PyTypeObject *> np.ndarray,
>>>>>>> np.dtype('<B'), 2,dims, NULL,pBuf, np.NPY_C_CONTIGUOUS, None)
>>>>>>
>>>>>> #-----------------------------------------------------------
>>>>>>> ------------------------------------------------
>>>>>>
>>>>>> #method 2:
>>>>>>
>>>>>> #if (i==0) or (i == 1) or (i ==2) or (i==3) or (i==4):
>>>>>>
>>>>>> #_puc_image = pBuf
>>>>>>
>>>>>> #pus_image = <unsigned short*>pBuf
>>>>>>
>>>>>> #for j in range(self.aoiWidth/self.hbin):
>>>>>>
>>>>>> #pus_image = <unsigned short*>(_puc_image)
>>>>>>
>>>>>> #for k in range(self.aoiHeight/self.vbin):
>>>>>>
>>>>>> #data[l][j][k]  = pus_image[0]
>>>>>>
>>>>>> #pus_image += 1
>>>>>>
>>>>>> #_puc_image += aoiStride
>>>>>>
>>>>>> #l = l+1
>>>>>>
>>>>>> the image displayed using method 1 is very different from method 1.
>>>>>
>>>>> ᐧ
>>>>>
>>>>> On Thu, May 24, 2018 at 9:34 AM, aishwarya selvaraj <
>>>>> aishwaryaselvaraj1708 at gmail.com> wrote:
>>>>>
>>>>>> Hi, Yes I was able to make use of  PyArray_NewFromDescr in my code.
>>>>>> Storing into a 3D array has now become a lot faster.Thank you!
>>>>>> But the CCD images stored using this method looks quite different
>>>>>> from the ones acquired earlier. I'm yet to find the root of the problem for
>>>>>> the same. Could you please take a look at it? I'm not very sure what
>>>>>> am I missing here.
>>>>>> ᐧ
>>>>>>
>>>>>> On Wed, May 23, 2018 at 7:43 AM, Aravindhan K <
>>>>>> aravindhank1994 at gmail.com> wrote:
>>>>>>
>>>>>>> Hi,
>>>>>>> Please use this example code. https://pastebin.com/nh9rvqfm.
>>>>>>> I have used this link as a reference.
>>>>>>> https://groups.google.com/forum/#!topic/cython-users/fFBiCCwblD0
>>>>>>>
>>>>>>>
>>>>>>> Regards,
>>>>>>> Aravindhan K
>>>>>>>
>>>>>>> On Tue, May 22, 2018 at 11:58 AM, aishwarya selvaraj <
>>>>>>> aishwaryaselvaraj1708 at gmail.com> wrote:
>>>>>>>
>>>>>>>> Hi,
>>>>>>>> Thank you for the suggestion. I was trying out an example code
>>>>>>>> using  PyArray_NewFromDescr*.* PFA of the code.I compiled the file
>>>>>>>> using :python setup.py build_ext --inplace.
>>>>>>>> When compiled I get the following error :
>>>>>>>> fatal error : numpy/arrayobject.h :No such file or directory.
>>>>>>>> I would be realy gratefull if you could do the edit in the file I
>>>>>>>> sent in my first mail.
>>>>>>>>
>>>>>>>> ᐧ
>>>>>>>>
>>>>>>>> On Mon, May 21, 2018 at 9:19 PM, Aravindhan K <
>>>>>>>> aravindhank1994 at gmail.com> wrote:
>>>>>>>>
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>> You could try using PyArray_NewFromDescr NumPy C-API function for
>>>>>>>>> creating nd-array
>>>>>>>>> in function capturing frames.
>>>>>>>>> than doing nested loops to do copy byte by byte. (This could be
>>>>>>>>> the bottleneck)
>>>>>>>>> https://docs.scipy.org/doc/numpy-1.14.0/reference/c-api.arra
>>>>>>>>> y.html#from-scratch
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Regards,
>>>>>>>>> Aravindhan K
>>>>>>>>>
>>>>>>>>> On Mon, May 21, 2018 at 5:12 PM, aishwarya selvaraj <
>>>>>>>>> aishwaryaselvaraj1708 at gmail.com> wrote:
>>>>>>>>>
>>>>>>>>>> Dear all,
>>>>>>>>>> I was programming for Andor CCD using SDK libraries in python
>>>>>>>>>> using cython.
>>>>>>>>>> I have a piece of code within my .pyx script which does the
>>>>>>>>>> function of storing the data into a 3D array. While acquiring n frames,
>>>>>>>>>> each of them is stored in a 3D array named data, which later I would be
>>>>>>>>>> displaying. But this creates a bottleneck in my code.
>>>>>>>>>> Hence I need to introduce threading, such that for each frame to
>>>>>>>>>> be stored there are individual threads for the same and the frames could be
>>>>>>>>>> displayed simultaneously without creating a performance lag.
>>>>>>>>>> Since I have never worked on multithreading, could anyone please
>>>>>>>>>> help me.
>>>>>>>>>> PFA of the code. (LiveAcquisition code)
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> --
>>>>>>>>>> Regards,
>>>>>>>>>> Aishwarya Selvaraj
>>>>>>>>>>
>>>>>>>>>> _______________________________________________
>>>>>>>>>> Chennaipy mailing list
>>>>>>>>>> Chennaipy at python.org
>>>>>>>>>> https://mail.python.org/mailman/listinfo/chennaipy
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> Chennaipy mailing list
>>>>>>>>> Chennaipy at python.org
>>>>>>>>> https://mail.python.org/mailman/listinfo/chennaipy
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> --
>>>>>>>> Regards,
>>>>>>>> Aishwarya Selvaraj
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> Chennaipy mailing list
>>>>>>>> Chennaipy at python.org
>>>>>>>> https://mail.python.org/mailman/listinfo/chennaipy
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Chennaipy mailing list
>>>>>>> Chennaipy at python.org
>>>>>>> https://mail.python.org/mailman/listinfo/chennaipy
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> Regards,
>>>>>> Aishwarya Selvaraj
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Regards,
>>>>> Aishwarya Selvaraj
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> Regards,
>>>> Aishwarya Selvaraj
>>>>
>>>> _______________________________________________
>>>> Chennaipy mailing list
>>>> Chennaipy at python.org
>>>> https://mail.python.org/mailman/listinfo/chennaipy
>>>>
>>>>
>>>
>>> _______________________________________________
>>> Chennaipy mailing list
>>> Chennaipy at python.org
>>> https://mail.python.org/mailman/listinfo/chennaipy
>>>
>>>
>>
>>
>> --
>> Regards,
>> Aishwarya Selvaraj
>>
>> _______________________________________________
>> Chennaipy mailing list
>> Chennaipy at python.org
>> https://mail.python.org/mailman/listinfo/chennaipy
>>
>>
>
> _______________________________________________
> Chennaipy mailing list
> Chennaipy at python.org
> https://mail.python.org/mailman/listinfo/chennaipy
>
>


-- 
Regards,
Aishwarya Selvaraj
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/chennaipy/attachments/20180603/b1dc75ac/attachment-0001.html>

More information about the Chennaipy mailing list