GPU install fails about yolo3-4-py HOT 3 CLOSED

The log indicates that compiling of Darknet has failed. (It is part of the installation process of yolo34py). It seems there is an issue with nVidia CUDA toolkit installation. The following line suggests that nvcc (nvidia CUDA compiler) cannot find gcc (GNU C compiler).

ERROR: No supported gcc/g++ host compiler found, but clang-3.8 is available.
           Use 'nvcc -ccbin clang-3.8' to use that instead.

Do you have gcc installed in your machine? You need to have a version of gcc compatible with the version of nVidia Toolkit installed. (So maybe you need to update your gcc)

from yolo3-4-py.

Yeah. Sorry, this was entirely my fault. I got frustrated and did some bad things to the packages on my system.

The darknet built itself failed with:

  ERROR: No supported gcc/g++ host compiler found, but clang-3.8 is available.
         Use 'nvcc -ccbin clang-3.8' to use that instead.
  make: *** [obj/convolutional_kernels.o] Error 1

On Debian Stretch, the only way I can find around this is to build Darknet from source and manually add -ccbin clang-3.8 to the NVCC command in the Makefile (in addition to setting GPU=1), and then install with DARKNET_HOME=/path/to/darknet LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$DARKNET_HOME pip3 install yolo34py-gpu

from yolo3-4-py.

Great that you managed to find a solution. Package issues can become very tricky!
Cheers and Good Luck!

from yolo3-4-py.