WARNING: The NVIDIA Driver was not detected

Robotics & Edge Computing Jetson Systems Jetson Thor
1 
Hi, I am testing my Jetson AGX Thor development kit I just bought. I have installed the Jetson ISO image by using an installation USB and also installed Jetpack 7.When I make the Docker Setup Test by running a PyTorch container as the 
following command:
"docker run --rm -it \
    -v "$PWD":/workspace \
    -w /workspace \
    nvcr.io/nvidia/pytorch:25.08-py3"
The terminal displayed errors as: 
"WARNING: The NVIDIA Driver was not detected.  GPU functionality will not be available.
   Use the NVIDIA Container Toolkit to start this container with GPU support; see
   https://docs.nvidia.com/datacenter/cloud-native/ ."

So, my questions are:
1. Is the GPU driver not installed when the Jetson ISO image was installed?
2. Do I need to reinstall the GPU driver? How to install the driver and what is the compatible version of the driver?  Thanks. BEN
2

You could to ‘sudo apt search nvidia-container’ to make sure it is installed and if not

sudo apt install nvidia-container

and make sure this file /etc/docker/daemon.json has these contents.

{
    "runtimes": {
        "nvidia": {
            "args": [],
            "path": "nvidia-container-runtime"
        }
    },
    "default-runtime": "nvidia"
}

and or add this to your docker run --runtime nvidia

3

Hi, whitesscott,

Thank you for your reply. Now the GPU driver is detected. Then I continue to perform CUDA setup according to the user guide.
I are trying the Example 2: Build cuda-samples using NGC CUDA container, and input the following command:
“cd ~
mkdir -p $HOME/cuda-work && cd $HOME/cuda-work
docker run --rm -it
-v “$PWD”:/workspace
-w /workspace
nvcr.io/nvidia/cuda:13.0.0-devel-ubuntu24.04”

After the CUDA container is running, then I input:
“apt update && apt install -y --no-install-recommends git make cmake
git clone --depth=1 --branch v13.0 GitHub - NVIDIA/cuda-samples: Samples for CUDA Developers which demonstrates features in CUDA Toolkit
cd cuda-samples/Samples/1_Utilities/deviceQuery
cmake . -DGPU_TARGETS=all -DCMAKE_BUILD_TYPE=Release
make -j$(nproc)
./deviceQuery”

But the result displays failure:
"root@eea659b120f8:/workspace/cuda-samples/Samples/1_Utilities/deviceQuery# ./deviceQuery
./deviceQuery Starting


CUDA Device Query (Runtime API) version (CUDART static linking)
cudaGetDeviceCount returned 803
→ system has unsupported display driver / cuda driver combination
Result = FAIL"

So, can you tell what wrong with this? Why “system has unsupported display driver / cuda driver combination”?

Thanks.

4

I tried the sudo apt install nvidia-cuda-samples and the build failed; those are not for Thor but Orin I think. I then decided to try v13.0 which was just recently released and has been modified to work with Thor / CUDA 13.

Here’s the steps I took to install them natively on Thor. You could do the same presumably in a container.

sudo apt install cmake openmpi-bin libopenmpi-dev

wget https://github.com/NVIDIA/cuda-samples/archive/refs/tags/v13.0.tar.gz
tar xfz v13.0.tar.gz
rm v13.0.tar.gz

cd cuda-samples-13.0
mkdir build && cd build
cmake ..

make SMS="110" -j12

# There are many "warning" message while compiling but they do not effect the built executables.

# The sms=110 tells compiler to optimize the build for Thor. -j12 is # of processors to use.

# Then I moved the cuda-samples-13.0/build/Samples to home.

mv ./Samples ~/

The executables are in the various sub directories of Samples and can be run from home or from in the subdir.

~/Samples/1_Utilities/deviceQuery/deviceQuery
/home/scott/Samples/1_Utilities/deviceQuery/deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 1 CUDA Capable device(s)

Device 0: "NVIDIA Thor"
  CUDA Driver Version / Runtime Version          13.0 / 13.0
  CUDA Capability Major/Minor version number:    11.0
  Total amount of global memory:                 125772 MBytes (131881758720 bytes)
  (020) Multiprocessors, (128) CUDA Cores/MP:    2560 CUDA Cores
  GPU Max Clock rate:                            1049 MHz (1.05 GHz)
  Memory Clock rate:                             0 Mhz
  Memory Bus Width:                              0-bit
  L2 Cache Size:                                 33554432 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total shared memory per multiprocessor:        233472 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  1536
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 1 copy engine(s)
  Run time limit on kernels:                     Yes
  Integrated GPU sharing Host Memory:            Yes
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Managed Memory:                Yes
  Device supports Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 1 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 13.0, CUDA Runtime Version = 13.0, NumDevs = 1
Result = PASS
5

Hi, whitesscott,

I just tried as your suggestion in the CUDA container, the problem is still the same:

“./deviceQuery Starting


CUDA Device Query (Runtime API) version (CUDART static linking)

cudaGetDeviceCount returned 803
→ system has unsupported display driver / cuda driver combination
Result = FAIL”

So, I do not think the problem is related to the release. It is already V13.0.

Can you try this in the nvcr.io/nvidia/cuda:13.0.0-devel-ubuntu24.04 container on the Jetson AGX thor development kit?

6

Hi, whitesscott,

I just install Jetpack 7, and I perform your suggested operations natively(not in container), it succeeds. So, I think the problem arise in cuda container, but not arise natively.

You can try this in cuda container, and you can repeat this problem and find a way to solve it. If you find the way to solve it in container, please share it with me.
Thanks. BEN

7

try this

cd ~/Samples  #or where you have your natively compiled cuda-samples/Samples

docker run -it --net=host --runtime nvidia --privileged --ipc=host --ulimit memlock=-1 \
 --ulimit stack=67108864 -v $(pwd):/workspace nvcr.io/nvidia/pytorch:25.08-py3 bash
8

Hi, whitesscott, I just tried as you said, and it succeeds. Thanks. BEN