Unable to install cuda-toolkit-12-6 inside 24.04 container

I am trying to install cuda-toolkit-12-6 inside the dustynv container:
dustynv/ros:jazzy-desktop-r36.4.0-cu128-24.04, however it is failing with the following:

190.5 Errors were encountered while processing:
190.5  /tmp/apt-dpkg-install-kzQ5kX/19-cuda-cudart-12-8_12.8.57-1_arm64.deb
190.5  /tmp/apt-dpkg-install-kzQ5kX/20-cuda-driver-dev-12-8_12.8.57-1_arm64.deb
190.5  /tmp/apt-dpkg-install-kzQ5kX/21-cuda-cudart-dev-12-8_12.8.57-1_arm64.deb
191.0 E: Sub-process /usr/bin/dpkg returned an error code (1)
------
Dockerfile:6
--------------------
   4 |     
   5 |     RUN /bin/bash -c "wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/sbsa/cuda-keyring_1.1-1_all.deb && dpkg -i cuda-keyring_1.1-1_all.deb"
   6 | >>> RUN apt-get update && apt-get -y --no-install-recommends install cuda-toolkit-12-6
   7 |     RUN apt-get update && apt-get -y --no-install-recommends install tensorrt
   8 |     

the host is an Orin AGX on 22.04 with cuda12.6 installed. The full docker file is:

FROM dustynv/ros:jazzy-desktop-r36.4.0-cu128-24.04
RUN apt-get update && apt-get -y --no-install-recommends install nano nlohmann-json3-dev python3-pip cmake libopenblas-dev

RUN /bin/bash -c "wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/sbsa/cuda-keyring_1.1-1_all.deb && dpkg -i cuda-keyring_1.1-1_all.deb"
RUN apt-get update && apt-get -y --no-install-recommends install cuda-toolkit-12-6
RUN apt-get update && apt-get -y --no-install-recommends install tensorrt

fundamentally i dont understand why the 3 following debs are looked for since they are 12.8 and this install is for 12.6:

190.5  /tmp/apt-dpkg-install-kzQ5kX/19-cuda-cudart-12-8_12.8.57-1_arm64.deb
190.5  /tmp/apt-dpkg-install-kzQ5kX/20-cuda-driver-dev-12-8_12.8.57-1_arm64.deb
190.5  /tmp/apt-dpkg-install-kzQ5kX/21-cuda-cudart-dev-12-8_12.8.57-1_arm64.deb

Hi,

It looks like the sbsa packages are used.

SBSA is an arm system with dGPU (arm64-sbsa) and is different from the Jetson which is an iGPU environment (aarch64-jetson).
Please try the package in the below link instead:

Thanks.

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