[Bug]: dlib.DLIB_USE_CUDA: false

[huan415]

What Operating System(s) are you seeing this problem on?

Linux (x86-64)

dlib version

19.14.0

Python version

3.8.10

Compiler

MSVC 19

Expected Behavior

import dlib
dlib.DLIB_USE_CUDA returns TRUE

Current Behavior

import dlib
dlib.DLIB_USE_CUDA returns FALSE

Steps to Reproduce

Dockerfile:

FROM nvidia/cuda:11.4.0-cudnn8-devel-ubuntu20.04
.....
RUN cd dlib-19.14.0 &&
mkdir build &&
cd build &&
cmake .. -DDLIB_USE_CUDA=1 -DUSE_AVX_INSTRUCTIONS=1 &&
cmake --build .
RUN cd dlib-19.14.0 &&
python3 setup.py install --yes USE_AVX_INSTRUCTIONS --yes DLIB_USE_CUDA

run

docker run --gpus all ...

env

nivdia-smi、nvcc -V is success. but dlib.DLIB_USE_CUDA returns false

Anything else?

No response

[huan415]

CUDA was found but your compiler failed to compile a simple CUDA program so dlib isn't going to use CUDA.

[arrufat]

I guess the most intriguing part is how you can use MSVC 19 on Linux.

[huan415]

MSVC

I guess the most intriguing part is how you can use MSVC 19 on Linux.

gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-44)