Unable to run sample hello world app on orin

Sistla3 · February 7, 2023, 10:36pm

Please provide the following info (tick the boxes after creating this topic):
Software Version
DRIVE OS 6.0.5
DRIVE OS 6.0.4 (rev. 1)
DRIVE OS 6.0.4 SDK
other

Target Operating System
Linux
QNX
other

Hardware Platform
DRIVE AGX Orin Developer Kit (940-63710-0010-D00)
DRIVE AGX Orin Developer Kit (940-63710-0010-200)
DRIVE AGX Orin Developer Kit (not sure its number)
other

SDK Manager Version
1.9.1.10844
other

Host Machine Version
native Ubuntu Linux 20.04 Host installed with SDK Manager
native Ubuntu Linux 20.04 Host installed with DRIVE OS Docker Containers
native Ubuntu Linux 18.04 Host installed with DRIVE OS Docker Containers
other

Hi I am using nvidia driver os docker container and i am trying to run the sample hello world application. I see the below error

root@6.0.5.0-0003-build-linux-sdk:/usr/local/driveworks/bin# ./sample_hello_world

Welcome to Driveworks SDK
[07-02-2023 22:04:40] Platform: Detected Generic x86 Platform
[07-02-2023 22:04:40] TimeSource: monotonic epoch time offset is 1675804657534688
[07-02-2023 22:04:40] Adding variable DW_Base:DW_Version
[07-02-2023 22:04:40] Addded variable DW_Base:DW_Version
[07-02-2023 22:04:40] Driveworks exception thrown: Platform: cannot acquire CUDA context. Error cudaErrorSystemDriverMismatch: system has unsupported display driver / cuda driver combination

Cannot init SDK
root@6.0.5.0-0003-build-linux-sdk:/usr/local/driveworks/bin#

On my host i installed the gpu driver and sdkmanager already.

±----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 1303 G /usr/lib/xorg/Xorg 54MiB |
| 0 N/A N/A 2204 G /usr/bin/gnome-shell 74MiB |
±----------------------------------------------------------------------------+

Is it expected that we need to install the same driver and sdkmanager in the docker environment as well to mitigate this issue?

Sistla3 · February 8, 2023, 12:25pm

I tried checking for the cuda driver on the host (not in the docker though) and i see the following :

nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Sun_Jul_28_19:07:16_PDT_2019
Cuda compilation tools, release 10.1, V10.1.243

SivaRamaKrishnaNV · February 8, 2023, 12:43pm

Dear @Sistla3,
Yes. It looks like issue with nvidia driver.
Did you install CUDA 12.x seperately on host? Note that, we recommend to use docker container for cross compiling. If you want to test out x86 samples, You can install DRIVE OS 6.0.5 using sdkmanager on host and check running.

Sistla3 · February 8, 2023, 1:34pm

Hi Siva,
The issue that i am noticing above is in the docker itself . I am not planning to run anything on the host side. But wanted to check if the issue is because of the driver missing on the host side

Sistla3 · February 8, 2023, 2:26pm

±----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 1396 G /usr/lib/xorg/Xorg 85MiB |
±----------------------------------------------------------------------------+

and i still see the same error on docker

root@6.0.5.0-0003-build-linux-sdk:/drive# /usr/local/driveworks/bin/sample_hello_world

Welcome to Driveworks SDK
[08-02-2023 14:25:00] Platform: Detected Generic x86 Platform
[08-02-2023 14:25:00] TimeSource: monotonic epoch time offset is 1675866044535784
[08-02-2023 14:25:00] Adding variable DW_Base:DW_Version
[08-02-2023 14:25:00] Addded variable DW_Base:DW_Version
[08-02-2023 14:25:00] Driveworks exception thrown: Platform: cannot acquire CUDA context. Error cudaErrorSystemDriverMismatch: system has unsupported display driver / cuda driver combination

Cannot init SDK
root@6.0.5.0-0003-build-linux-sdk:/drive#

Is there any procedure document that needs to be followed for the tools installations on the host side before starting to use the docker?
is there a way for me to check the driver version that is being used in the docker to rule out any mismatches if there are any ?
I tried checking the nvidia-smi inside the docker and i dont see any installed binary for that and it fails to tell me what the driver version in the container is.
Will a lightdm on the host machine cause any of these errors?

Sistla3 · February 9, 2023, 8:07pm

Hi Nvidia ,
Can you please add some inputs on the problem that i am seeing here. I am still stuck with this.

Regards,
Sistla.

Sistla3 · February 10, 2023, 3:09am

Ok for any one visiting this page here is the solution for this issue :

This issue happens if the nvidia docker cannot recognize the underlying nvidia drivers.

To resolve the issue i followed the page
https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html
I installed the CTK and made sure to follow the instructions to make sure that the visible into the docker.

make sure to restart the daemon at end.

here is the docker run command that i used

sudo docker run --rm --runtime=nvidia --gpus all -it --privileged --net=host -v /dev/bus/usb:/dev/bus/usb -v /home/autonav23/docker_share:/home/nvidia/ nvcr.io/drive/driveos-sdk/drive-agx-orin-linux-aarch64-sdk-build-x86:latest

system · February 24, 2023, 3:09am

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.