Cuda:13.2.1 runtime image can't run on r39.2 sample rootfs for Orin Nano

Robotics & Edge Computing Jetson Systems Jetson Orin Nano
1

I am currently using a Jetson Orin Nano 8GB with BSP R39.2 and the official sample rootfs. After pulling the image, I can’t run this image successfully. But if it can be work for image: nvcr.io/nvidia/l4t-jetpack:r36.3.0. Please help to check this issue, thanks.

sudo docker pull --platform linux/arm64 nvcr.io/nvidia/cuda:13.2.1-runtime-ubuntu24.04

kuka@kuka:~$ sudo docker run --rm -itd --net=host --runtime nvidia -v /home/kuka/userdata:/root/userdata nvcr.io/nvidia/cuda:13.2.1-runtime-ubuntu24.04
e880013a50c1f31efb06a63db45d454e3fba3517779673fcce270a066bf8dddd
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running createContainer hook #2: exit status 2, stdout: , stderr: panic: runtime error: slice bounds out of range [:73] with capacity 71

goroutine 1 [running]:
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.trim(...)
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cuda-elf-header.go:100
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.GetCUDACompatElfHeaderFromReader({0x97af80?, 0x62927200640?})
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cuda-elf-header.go:86 +0x538
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.command.useCompatLibraries({{0x97e020?, 0x629272d1500?}}, 0x2a?, {0x0?, 0x6292722abe8?}, {0xffffe2ef7f8f, 0x4})
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cudacompat.go:197 +0x334
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.command.getContainerForwardCompatDir({{0x97e020?, 0x629272d1500?}}, 0x629272005c0, 0x6292727ee40)
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cudacompat.go:180 +0x2b4
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.command.run({{0x97e020?, 0x629272d1500?}}, 0x629273858c8?, 0x6292727ee40)
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cudacompat.go:135 +0x1fc
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.command.build.func2({0x629273a2848?, 0x97d4b0?}, 0x62927393980?)
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cudacompat.go:76 +0x24
github.com/urfave/cli/v3.(*Command).run(0x629273a2848, {0x97d4b0, 0x62927393980}, {0x62927393890, 0x3, 0x3})
        /go/src/nvidia-container-toolkit/vendor/github.com/urfave/cli/v3/command_run.go:360 +0x2214
github.com/urfave/cli/v3.(*Command).run(0x62927385b88, {0x97d4b0, 0x629273937a0}, {0x6292727f180, 0x4, 0x4})
        /go/src/nvidia-container-toolkit/vendor/github.com/urfave/cli/v3/command_run.go:294 +0x1c4c
github.com/urfave/cli/v3.(*Command).run(0x629273858c8, {0x97d4b0, 0x629273929c0}, {0x62927222050, 0x5, 0x5})
        /go/src/nvidia-container-toolkit/vendor/github.com/urfave/cli/v3/command_run.go:294 +0x1c4c
github.com/urfave/cli/v3.(*Command).Run(...)
        /go/src/nvidia-container-toolkit/vendor/github.com/urfave/cli/v3/command_run.go:93
main.main()
        /go/src/nvidia-container-toolkit/cmd/nvidia-ctk/main.go:101 +0x7a8
2

— 🤖 This is an automated AI response. I’m here to help, but please verify important details! —

I could not find clear support for that in the provided Jetson Orin Nano documentation. The error message you’re encountering seems related to compatibility issues between the CUDA runtime image and the sample rootfs for Orin Nano with BSP R39.2. However, the provided context does not specifically address this issue or provide a solution for running the cuda:13.2.1-runtime-ubuntu24.04 image on the Jetson Orin Nano with the mentioned configuration. For further assistance, I recommend checking the official NVIDIA forums or documentation for the most up-to-date information and troubleshooting guides related to your specific setup and version of JetPack and L4T.

— We support team will also look into your issue to have further suggestions soon. please stay tuned. —

4

Hi,

Please try:

$ docker pull nvcr.io/nvidia/cuda:13.3.0-runtime-ubuntu24.04
$ docker run --gpus all -it --rm nvcr.io/nvidia/cuda:13.3.0-runtime-ubuntu24.04

Thanks.

5

Hi,

I try to use 13.3.0 image with --gpus all and --runtime=nvidia, but it can’t work ether. Please check the log below, thanks.

kuka@kuka:~$ sudo docker run --gpus all -it --rm nvcr.io/nvidia/cuda:13.3.0-runtime-ubuntu24.04
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running prestart hook #0: exit status 1, stdout: , stderr: Auto-detected mode as 'csv'
invoking the NVIDIA Container Runtime Hook directly (e.g. specifying the docker --gpus flag) is not supported. Please use the NVIDIA Container Runtime (e.g. specify the --runtime=nvidia flag) instead

Run 'docker run --help' for more information

kuka@kuka:~$ sudo docker run --runtime=nvidia -it --rm nvcr.io/nvidia/cuda:13.3.0-runtime-ubuntu24.04
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: could not apply required modification to OCI specification: error modifying OCI spec: requirements not met: cuda>=13.3||brand=unknown&&driver>=535&&driver<536||brand=grid&&driver>=535&&driver<536||brand=tesla&&driver>=535&&driver<536||brand=nvidia&&driver>=535&&driver<536||brand=quadro&&driver>=535&&driver<536||brand=quadrortx&&driver>=535&&driver<536||brand=nvidiartx&&driver>=535&&driver<536||brand=vapps&&driver>=535&&driver<536||brand=vpc&&driver>=535&&driver<536||brand=vcs&&driver>=535&&driver<536||brand=vws&&driver>=535&&driver<536||brand=cloudgaming&&driver>=535&&driver<536||brand=unknown&&driver>=550&&driver<551||brand=grid&&driver>=550&&driver<551||brand=tesla&&driver>=550&&driver<551||brand=nvidia&&driver>=550&&driver<551||brand=quadro&&driver>=550&&driver<551||brand=quadrortx&&driver>=550&&driver<551||brand=nvidiartx&&driver>=550&&driver<551||brand=vapps&&driver>=550&&driver<551||brand=vpc&&driver>=550&&driver<551||brand=vcs&&driver>=550&&driver<551||brand=vws&&driver>=550&&driver<551||brand=cloudgaming&&driver>=550&&driver<551||brand=unknown&&driver>=560&&driver<561||brand=grid&&driver>=560&&driver<561||brand=tesla&&driver>=560&&driver<561||brand=nvidia&&driver>=560&&driver<561||brand=quadro&&driver>=560&&driver<561||brand=quadrortx&&driver>=560&&driver<561||brand=nvidiartx&&driver>=560&&driver<561||brand=vapps&&driver>=560&&driver<561||brand=vpc&&driver>=560&&driver<561||brand=vcs&&driver>=560&&driver<561||brand=vws&&driver>=560&&driver<561||brand=cloudgaming&&driver>=560&&driver<561||brand=unknown&&driver>=565&&driver<566||brand=grid&&driver>=565&&driver<566||brand=tesla&&driver>=565&&driver<566||brand=nvidia&&driver>=565&&driver<566||brand=quadro&&driver>=565&&driver<566||brand=quadrortx&&driver>=565&&driver<566||brand=nvidiartx&&driver>=565&&driver<566||brand=vapps&&driver>=565&&driver<566||brand=vpc&&driver>=565&&driver<566||brand=vcs&&driver>=565&&driver<566||brand=vws&&driver>=565&&driver<566||brand=cloudgaming&&driver>=565&&driver<566||brand=unknown&&driver>=570&&driver<571||brand=grid&&driver>=570&&driver<571||brand=tesla&&driver>=570&&driver<571||brand=nvidia&&driver>=570&&driver<571||brand=quadro&&driver>=570&&driver<571||brand=quadrortx&&driver>=570&&driver<571||brand=nvidiartx&&driver>=570&&driver<571||brand=vapps&&driver>=570&&driver<571||brand=vpc&&driver>=570&&driver<571||brand=vcs&&driver>=570&&driver<571||brand=vws&&driver>=570&&driver<571||brand=cloudgaming&&driver>=570&&driver<571||brand=unknown&&driver>=580&&driver<581||brand=grid&&driver>=580&&driver<581||brand=tesla&&driver>=580&&driver<581||brand=nvidia&&driver>=580&&driver<581||brand=quadro&&driver>=580&&driver<581||brand=quadrortx&&driver>=580&&driver<581||brand=nvidiartx&&driver>=580&&driver<581||brand=vapps&&driver>=580&&driver<581||brand=vpc&&driver>=580&&driver<581||brand=vcs&&driver>=580&&driver<581||brand=vws&&driver>=580&&driver<581||brand=cloudgaming&&driver>=580&&driver<581 not met

I try to run static link deviceQuery on host side. It’s worked. So I think the driver is ok.

kuka@kuka:~$ ./deviceQuery
./deviceQuery Starting...

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

Detected 1 CUDA Capable device(s)

Device 0: "Orin"
  CUDA Driver Version / Runtime Version          13.2 / 10.2
  CUDA Capability Major/Minor version number:    8.7
  Total amount of global memory:                 7547 MBytes (7913598976 bytes)
MapSMtoCores for SM 8.7 is undefined.  Default to use 64 Cores/SM
MapSMtoCores for SM 8.7 is undefined.  Default to use 64 Cores/SM
  ( 8) Multiprocessors, ( 64) CUDA Cores/MP:     512 CUDA Cores
  GPU Max Clock rate:                            624 MHz (0.62 GHz)
  Memory Clock rate:                             624 Mhz
  Memory Bus Width:                              128-bit
  L2 Cache Size:                                 2097152 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 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 2 copy engine(s)
  Run time limit on kernels:                     No
  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 Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 0 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 13.2, CUDA Runtime Version = 10.2, NumDevs = 1
Result = PASS

I also try to use 13.2.0 image, but can’t work ether.

kuka@kuka:~$ sudo docker run --runtime=nvidia -it --rm nvcr.io/nvidia/cuda:13.2.0-runtime-ubuntu24.04
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running createContainer hook #2: exit status 2, stdout: , stderr: panic: runtime error: slice bounds out of range [:73] with capacity 71

goroutine 1 [running]:
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.trim(...)
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cuda-elf-header.go:100
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.GetCUDACompatElfHeaderFromReader({0x97af80?, 0x13931b1b4640?})
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cuda-elf-header.go:86 +0x538
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.command.useCompatLibraries({{0x97e020?, 0x13931b305500?}}, 0x2a?, {0x0?, 0x13931b1debe8?}, {0xffffe5623f8f, 0x4})
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cudacompat.go:197 +0x334
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.command.getContainerForwardCompatDir({{0x97e020?, 0x13931b305500?}}, 0x13931b1b45c0, 0x13931b232e40)
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cudacompat.go:180 +0x2b4
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.command.run({{0x97e020?, 0x13931b305500?}}, 0x13931b3398c8?, 0x13931b232e40)
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cudacompat.go:135 +0x1fc
github.com/NVIDIA/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat.command.build.func2({0x13931b356848?, 0x97d4b0?}, 0x13931b347980?)
        /go/src/nvidia-container-toolkit/cmd/nvidia-cdi-hook/cudacompat/cudacompat.go:76 +0x24
github.com/urfave/cli/v3.(*Command).run(0x13931b356848, {0x97d4b0, 0x13931b347980}, {0x13931b347890, 0x3, 0x3})
        /go/src/nvidia-container-toolkit/vendor/github.com/urfave/cli/v3/command_run.go:360 +0x2214
github.com/urfave/cli/v3.(*Command).run(0x13931b339b88, {0x97d4b0, 0x13931b3477a0}, {0x13931b233180, 0x4, 0x4})
        /go/src/nvidia-container-toolkit/vendor/github.com/urfave/cli/v3/command_run.go:294 +0x1c4c
github.com/urfave/cli/v3.(*Command).run(0x13931b3398c8, {0x97d4b0, 0x13931b3469c0}, {0x13931b1d6050, 0x5, 0x5})
        /go/src/nvidia-container-toolkit/vendor/github.com/urfave/cli/v3/command_run.go:294 +0x1c4c
github.com/urfave/cli/v3.(*Command).Run(...)
        /go/src/nvidia-container-toolkit/vendor/github.com/urfave/cli/v3/command_run.go:93
main.main()
        /go/src/nvidia-container-toolkit/cmd/nvidia-ctk/main.go:101 +0x7a8

Run 'docker run --help' for more information
6

Hi,

It looks like there is a CUDA 10.2 library installed in your environment.
Could you remove it and check if you can load the 13.2 runtime as below?

./deviceQuery Starting...

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

Detected 1 CUDA Capable device(s)

Device 0: "Orin"
  CUDA Driver Version / Runtime Version          13.2 / 13.2
  CUDA Capability Major/Minor version number:    8.7
...

Thanks.

7

Hi

I check my cuda version I think it’s 13.2 already:

kuka@kuka:~$ sudo apt install cuda-13-2
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
cuda-13-2 is already the newest version (13.2.1-1).
0 upgraded, 0 newly installed, 0 to remove and 151 not upgraded.

And my deviceQuery is static link, so the runtime version it print out is not real runtime version in the system. I use rebuild cuda-samples, and run deviceQuery like this:

kuka@kuka:~/userdata/cuda-samples-13.2/build/Samples/1_Utilities/deviceQuery$ ./deviceQuery
./deviceQuery Starting...

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

Detected 1 CUDA Capable device(s)

Device 0: "Orin"
  CUDA Driver Version / Runtime Version          13.2 / 13.2
  CUDA Capability Major/Minor version number:    8.7
  Total amount of global memory:                 7547 MBytes (7913594880 bytes)
  (008) Multiprocessors, (128) CUDA Cores/MP:    1024 CUDA Cores
  GPU Max Clock rate:                            624 MHz (0.62 GHz)
  Memory Clock rate:                             624 Mhz
  Memory Bus Width:                              128-bit
  L2 Cache Size:                                 2097152 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:        167936 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 2 copy engine(s)
  Run time limit on kernels:                     No
  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 / 0 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 13.2, CUDA Runtime Version = 13.2, NumDevs = 1
Result = PASS


kuka@kuka:~$ dpkg -l | grep -i cuda
ii  cuda-13-2                                     13.2.1-1                                         arm64        CUDA 13.2 meta-package
ii  cuda-cccl-13-2                                13.2.75-1                                        arm64        CUDA CCCL
ii  cuda-command-line-tools-13-2                  13.2.1-1                                         arm64        CUDA command-line tools
ii  cuda-compiler-13-2                            13.2.1-1                                         arm64        CUDA compiler
ii  cuda-crt-13-2                                 13.2.78-1                                        arm64        CUDA crt
ii  cuda-cudart-13-2                              13.2.75-1                                        arm64        CUDA Runtime native Libraries
ii  cuda-cudart-dev-13-2                          13.2.75-1                                        arm64        CUDA Runtime native dev links, headers
ii  cuda-culibos-dev-13-2                         13.2.75-1                                        arm64        CUDA DEV culibos is a Math Libraries fork of the cuos library
ii  cuda-cuobjdump-13-2                           13.2.78-1                                        arm64        CUDA cuobjdump
ii  cuda-cupti-13-2                               13.2.75-1                                        arm64        CUDA profiling tools runtime libs.
ii  cuda-cupti-dev-13-2                           13.2.75-1                                        arm64        CUDA profiling tools interface.
ii  cuda-cuxxfilt-13-2                            13.2.78-1                                        arm64        CUDA cuxxfilt
ii  cuda-documentation-13-2                       13.2.75-1                                        arm64        CUDA documentation
ii  cuda-driver-dev-13-2                          13.2.75-1                                        arm64        CUDA Driver native dev stub library
ii  cuda-gdb-13-2                                 13.2.75-1                                        arm64        CUDA-GDB
ii  cuda-libraries-13-2                           13.2.1-1                                         arm64        CUDA Libraries 13.2 meta-package
ii  cuda-libraries-dev-13-2                       13.2.1-1                                         arm64        CUDA Libraries 13.2 development meta-package
ii  cuda-nsight-compute-13-2                      13.2.1-1                                         arm64        NVIDIA Nsight Compute
ii  cuda-nsight-systems-13-2                      13.2.1-1                                         arm64        NVIDIA Nsight Systems
ii  cuda-nvcc-13-2                                13.2.78-1                                        arm64        CUDA nvcc
ii  cuda-nvdisasm-13-2                            13.2.78-1                                        arm64        CUDA disassembler
ii  cuda-nvml-dev-13-2                            13.2.82-1                                        arm64        NVML native dev links, headers
ii  cuda-nvprune-13-2                             13.2.78-1                                        arm64        CUDA nvprune
ii  cuda-nvrtc-13-2                               13.2.78-1                                        arm64        NVRTC native runtime libraries
ii  cuda-nvrtc-dev-13-2                           13.2.78-1                                        arm64        NVRTC native dev links, headers
ii  cuda-nvtx-13-2                                13.2.75-1                                        arm64        NVIDIA Tools Extension
ii  cuda-profiler-api-13-2                        13.2.75-1                                        arm64        CUDA Profiler API
ii  cuda-runtime-13-2                             13.2.1-1                                         arm64        CUDA Runtime 13.2 meta-package
ii  cuda-sanitizer-13-2                           13.2.76-1                                        arm64        CUDA Sanitizer
ii  cuda-tileiras-13-2                            13.2.78-1                                        arm64        JIT compilation feature for TileIR.
ii  cuda-toolkit-13-2                             13.2.1-1                                         arm64        CUDA Toolkit 13.2 meta-package
ii  cuda-toolkit-13-2-config-common               13.2.75-1                                        all          Common config package for CUDA Toolkit 13.2.
ii  cuda-toolkit-13-config-common                 13.2.75-1                                        all          Common config package for CUDA Toolkit 13.
ii  cuda-toolkit-config-common                    13.2.75-1                                        all          Common config package for CUDA Toolkit.
ii  cuda-tools-13-2                               13.2.1-1                                         arm64        CUDA Tools meta-package
ii  cuda-visual-tools-13-2                        13.2.1-1                                         arm64        CUDA visual tools
ii  libcufile-13-2                                1.17.1.22-1                                      arm64        Library for GPU Direct Storage with CUDA 13.2
ii  libcusolver-13-2                              12.2.0.1-1                                       arm64        CUDA solver native runtime libraries
ii  libcusolver-dev-13-2                          12.2.0.1-1                                       arm64        CUDA solver native dev links, headers
ii  libnvptxcompiler-13-2                         13.2.78-1                                        arm64        CUDA nvptxcompiler
ii  libnvvm-13-2                                  13.2.78-1                                        arm64        CUDA nvvm
ii  nvidia-l4t-cuda                               39.2.0-20260601141651                            arm64        NVIDIA CUDA Package
ii  nvidia-l4t-cuda-nvgpu                         39.2.0-20260601141651                            arm64        NVIDIA CUDA Package specific to nvgpu
ii  nvidia-l4t-cuda-openrm                        39.2.0-20260601141651                            arm64        NVIDIA CUDA Package specific to openrm
ii  nvidia-l4t-cuda-utils                         39.2.0-20260601141651                            arm64        NVIDIA CUDA utilities
8

Hi,

Based on the log, your CUDA is working now.
Thanks.

9

Not yet, it just can run on the host side. Docker start failed issue still exist.

10

Hi,

What kind of error in the Docker test now?

Suppose the Docker and nvidia-container-toolkit are pre-installed in JetPack 7.2.
Do you re-install or modify the package?

We can run the cuda:13.3.0-runtime container without issue.
Thanks.

11

Hi,

I executed the following steps in sequence:

  1. tar jxf Jetson_Linux_R39.2.0_aarch64.tbz2

  2. sudo tar jxpf Tegra_Linux_Sample-Root-Filesystem_R39.2.0_aarch64.tbz2 -C ./Linux_for_Tegra/rootfs/

  3. sudo ./apply_binaries.sh

  4. sudo ./tools/kernel_flash/l4t_initrd_flash.sh -p "--no-systemimg -c bootloader/generic/cfg/flash_t234_qspi.xml" --network usb0 jetson-orin-nano-devkit nvme0n1p1

  5. Edit /etc/docker/daemon.json like:

     {
"registry-mirrors": [],

"log-driver":"json-file",

"log-opts": {"max-size":"1m", "max-file":"1"},

"runtimes": {
"nvidia": {
"path": "nvidia-container-runtime",
"runtimeArgs": []

}
}
}

    and then sudo systemctl restart docker

  6. Login to the board, config the wifi and run:docker pull nvcr.io/nvidia/cuda:13.3.0-runtime-ubuntu24.04

  7. kuka@kuka:~$ sudo docker run --gpus all -it --rm nvcr.io/nvidia/cuda:13.3.0-runtime-ubuntu24.04
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running prestart hook #0: exit status 1, stdout: , stderr: Auto-detected mode as 'csv'
invoking the NVIDIA Container Runtime Hook directly (e.g. specifying the docker --gpus flag) is not supported. Please use the NVIDIA Container Runtime (e.g. specify the --runtime=nvidia flag) instead

Run 'docker run --help' for more information

kuka@kuka:~$ sudo docker run --runtime=nvidia -it --rm nvcr.io/nvidia/cuda:13.3.0-runtime-ubuntu24.04
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: could not apply required modification to OCI specification: error modifying OCI spec: requirements not met: cuda>=13.3||brand=unknown&&driver>=535&&driver<536||brand=grid&&driver>=535&&driver<536||brand=tesla&&driver>=535&&driver<536||brand=nvidia&&driver>=535&&driver<536||brand=quadro&&driver>=535&&driver<536||brand=quadrortx&&driver>=535&&driver<536||brand=nvidiartx&&driver>=535&&driver<536||brand=vapps&&driver>=535&&driver<536||brand=vpc&&driver>=535&&driver<536||brand=vcs&&driver>=535&&driver<536||brand=vws&&driver>=535&&driver<536||brand=cloudgaming&&driver>=535&&driver<536||brand=unknown&&driver>=550&&driver<551||brand=grid&&driver>=550&&driver<551||brand=tesla&&driver>=550&&driver<551||brand=nvidia&&driver>=550&&driver<551||brand=quadro&&driver>=550&&driver<551||brand=quadrortx&&driver>=550&&driver<551||brand=nvidiartx&&driver>=550&&driver<551||brand=vapps&&driver>=550&&driver<551||brand=vpc&&driver>=550&&driver<551||brand=vcs&&driver>=550&&driver<551||brand=vws&&driver>=550&&driver<551||brand=cloudgaming&&driver>=550&&driver<551||brand=unknown&&driver>=560&&driver<561||brand=grid&&driver>=560&&driver<561||brand=tesla&&driver>=560&&driver<561||brand=nvidia&&driver>=560&&driver<561||brand=quadro&&driver>=560&&driver<561||brand=quadrortx&&driver>=560&&driver<561||brand=nvidiartx&&driver>=560&&driver<561||brand=vapps&&driver>=560&&driver<561||brand=vpc&&driver>=560&&driver<561||brand=vcs&&driver>=560&&driver<561||brand=vws&&driver>=560&&driver<561||brand=cloudgaming&&driver>=560&&driver<561||brand=unknown&&driver>=565&&driver<566||brand=grid&&driver>=565&&driver<566||brand=tesla&&driver>=565&&driver<566||brand=nvidia&&driver>=565&&driver<566||brand=quadro&&driver>=565&&driver<566||brand=quadrortx&&driver>=565&&driver<566||brand=nvidiartx&&driver>=565&&driver<566||brand=vapps&&driver>=565&&driver<566||brand=vpc&&driver>=565&&driver<566||brand=vcs&&driver>=565&&driver<566||brand=vws&&driver>=565&&driver<566||brand=cloudgaming&&driver>=565&&driver<566||brand=unknown&&driver>=570&&driver<571||brand=grid&&driver>=570&&driver<571||brand=tesla&&driver>=570&&driver<571||brand=nvidia&&driver>=570&&driver<571||brand=quadro&&driver>=570&&driver<571||brand=quadrortx&&driver>=570&&driver<571||brand=nvidiartx&&driver>=570&&driver<571||brand=vapps&&driver>=570&&driver<571||brand=vpc&&driver>=570&&driver<571||brand=vcs&&driver>=570&&driver<571||brand=vws&&driver>=570&&driver<571||brand=cloudgaming&&driver>=570&&driver<571||brand=unknown&&driver>=580&&driver<581||brand=grid&&driver>=580&&driver<581||brand=tesla&&driver>=580&&driver<581||brand=nvidia&&driver>=580&&driver<581||brand=quadro&&driver>=580&&driver<581||brand=quadrortx&&driver>=580&&driver<581||brand=nvidiartx&&driver>=580&&driver<581||brand=vapps&&driver>=580&&driver<581||brand=vpc&&driver>=580&&driver<581||brand=vcs&&driver>=580&&driver<581||brand=vws&&driver>=580&&driver<581||brand=cloudgaming&&driver>=580&&driver<581 not met

Other than the steps above, I did not perform any additional actions, and I did not modify any deb packages or the filesystem.

In addition, as indicated by the message above:

invoking the NVIDIA Container Runtime Hook directly (e.g. specifying the docker --gpus flag) is not supported. Please use the NVIDIA Container Runtime (e.g. specify the --runtime=nvidia flag) instead

the --gpus option is not supported. I am therefore quite confused as to why it works on your side. Could it be due to a module difference? We are using the Orin Nano 8GB module. I’m sure that nvidia-container-toolkit has been install.

Thank you in advance for your help, and I look forward to your reply.

12

Hi,

Our environment is setup with SDKmanager.
We will try it with the manual flash and get back to you soon.

Thanks.

13

Hi, is there any update?

Thanks.

14

Hi,

Sorry for the late update.

Could you try to add -e NVIDIA_DISABLE_REQUIRE=true to see if this can help?

Thanks.

15

Hi, I found that the SDK installs containerd.io and docker-ce by default instead of docker.io. So after I replaced docker.io with containerd.io and docker-ce, I was able to run nvcr.io/nvidia/cuda:13.3.0-runtime-ubuntu24.04. However, I still cannot run nvcr.io/nvidia/cuda:13.2.1-runtime-ubuntu24.04.

Since our product is intended for mass production, we need to assess whether running cuda:13.3.0 on the r39.2 BSP introduces any risks.

16

Hi,

Please add the -e NVIDIA_DISABLE_REQUIRE=true flag and try it again.
This is a known issue where the r590/r595 is not in the list of allowed driver versions (up to 58x currently).

We will fix this issue in the upcoming 13.3 update.
Thanks.

18

Hi,

We can start the container normally and use CUDA with the following command:

sudo docker run -e NVIDIA_DISABLE_REQUIRE=true --runtime=nvidia -v /home/kuka/:/root/userdata -itd --rm nvcr.io/nvidia/cuda:13.3.0-runtime-ubuntu24.04

However, using the 13.2.1 image still fails.

I would like to understand the following two points:

  1. What is the purpose of the parameter -e NVIDIA_DISABLE_REQUIRE=true? Does it introduce any risks?

  2. The BSP information for r39.2 shows that the CUDA version is 13.2.1. In that case, is there any risk in using the image nvcr.io/nvidia/cuda:13.3.0-runtime-ubuntu24.04? Is this the recommended approach?

19

Hi,

You will need the newer container for the Orin support.
Orin with SBSA driver support is added in JetPack 7.2.

1.
The -e NVIDIA_DISABLE_REQUIRE=true flag will skip the driver version check.
We will add the corresponding driver to our upcoming container so you don’t need to manually set the flag anymore.

2.
No, the version within the container doesn’t need to match the JetPack version.

Thanks.

20

Hi,

Just to confirm, for now should we use the cuda:13.3.0 image with -e NVIDIA_DISABLE_REQUIRE=true?

We are currently following the JetPack 7.2 archive page:

According to that page, we are using:

  • cuDNN 9.20.0
  • TensorRT 10.16.2

If we move to the cuda:13.3.0 image, do we need to update the cuDNN and TensorRT versions inside the container as well? If so, could you let us know which cuDNN and TensorRT versions are recommended for CUDA 13.3.0?

Thanks.

21

Hi,

Currently, please run the CUDA 13.3 container with -e NVIDIA_DISABLE_REQUIRE=true.

The library within the container is independent of the system library.
You will need to manually install TensorRT and cuDNN is required.

Thanks.