Currently, CUDA Python loads NVRTC using standard system paths, for example on Linux (cf man ld.so) we do:
|
handle = dlfcn.dlopen('libnvrtc.so.12', dlfcn.RTLD_NOW) |
However, it means unless users purposely add
site-packages to
LD_LIBRARY_PATH, which is a frown-upon practice, the
nvidia-cuda-nvrtc-cuXX wheel (ex:
cu12) installed in the user environment will not be used. This issue is for tracking the status of this enhancement.
Currently, CUDA Python loads NVRTC using standard system paths, for example on Linux (cf
man ld.so) we do:cuda-python/cuda/_cuda/cnvrtc.pyx.in
Line 56 in 2be0aac
However, it means unless users purposely add
site-packagestoLD_LIBRARY_PATH, which is a frown-upon practice, thenvidia-cuda-nvrtc-cuXXwheel (ex: cu12) installed in the user environment will not be used. This issue is for tracking the status of this enhancement.