cpython-performance

to compare performance between c/c++ and python language compare cpython c/c++ and pypy performance gcc ccalc.c -o ccalc pi@raspberrypi:~/PROJECT/temp $ ./ccalc time diff:0.678562 second

for cpp/c++ compile g++ Ccalc.cpp -o Ccalc.a run pi@raspberrypi:~/PROJECT/temp $ ./Ccalc.a time diff:0.659825 second

compare c/c++,python3, pypy performance pi@raspberrypi:/PROJECT/temp $ ./ccalc time diff:0.678562 second pi@raspberrypi:/PROJECT/temp $ ^C pi@raspberrypi:/PROJECT/temp $ g++ Ccalc.cpp -o Ccalc.a pi@raspberrypi:/PROJECT/temp $ ./Ccalc.a time diff:0.659825 second pi@raspberrypi:/PROJECT/temp $ python3 calc.py the time diff is :36.714549 pi@raspberrypi:/PROJECT/temp $ pypy calc.py the time diff is :15.094903

generate .so lib 1 gcc -shared psdcod.c -o psd.so 2 using .so in cpyhton 1)import ctypes as C 2) if SERVER: self.so = C.CDLL('./psdcod.so') #else: #self.so = C.CDLL('psccod.dll')

     self.infrRGB = np.zeros([60, 80, 3], np.uint8)
    self.infr = np.zeros([60, 80, 1], np.uint16)

     def pgm2BGR2(self, img1):
    img =np.array(img1)
    self.so.pgm2BGR(img.ctypes.data_as(C.POINTER(C.c_ubyte)),self.infrRGB.ctypes.data_as(C.POINTER(C.c_ubyte)))
    newimg = cv.resize(self.infrRGB, (160, 120), interpolation=cv.INTER_NEAREST)
    #newimg = self.infrRGB
    return newimg