NPP within Python3 segfaults on malloc?

Hi,

I’m trying to replicate some code which used to be available at Best Open Source Mac Software 2022 and seems to be someone saved at cudacam/nppThresholding.py at master · mattsta/cudacam · GitHub .

I’m running CUDA 9.0 with a Titan V card on Linux Mint (Ubuntu). Other Python-CUDA applications (e.g. homebrew CuPy stuff) work perfectly fine, even impressively so, compared to their speed when I had a 1080Ti installed.

The reduced offending code is:

from ctypes import *
npp_supp = CDLL("/usr/local/cuda/lib64/libnppisu.so")
npp_supp.nppiMalloc_8u_C1.restype = POINTER(c_char)
d_a = npp_supp.nppiMalloc_8u_C1(640, 480, 640)

I’ll sure be grateful if someone can help me figure out what I’m doing wrong here!