Difference Between Linux/aarch64 and Jetson/aarch64

Robotics & Edge Computing Jetson Systems Jetson TX2
1

Hello,

I want to use Jetson aarch64 docker container on Windows but docker multiarchitecture does not support jetson/aarch64. İt only supports linux/aarch64. What is difference between linux/aarch64 and jetpack/aarxh64? If I use linux/aarch64 for container instead of jetson/aarch64, is there any major difference or it is ok? (Briefly I am curious about their difference)

4

Hi,

Could you share with us your use case and why you need to run the Jetson image on the host?

In general, the Jetson-based image indicates support for the integrated GPU.
Other aarch64 images might build for the discrete GPU.

If you want to cross-compile an app on the host, you can use the image below instead:

Thanks.

5

Hello @AastaLLL

I want to test my jetson device projects without using physically Jetson device. So I want to use Jetson container on my windows computer. That’s why I want to run Jetson image on Windows (x86_64)

If I use the link that you shared, it gives me the cross compile but can I test and run my project in this container on Windows?

6

Hi,

It depends on the use case.

For example, if you use PyTorch + OpenCV.
You can test it with a desktop container that installed the same PyTorch version.

But if your app use some Jetson-specific hardware, ex Deepstream or MMAPI.
Please test it directly on the Jetson device.

Thanks.

7

Hello @AastaLLL

I will use Pytorch and similar libraries, can I use the container that you shared with me (JetPack Cross Compilation container | NVIDIA NGC) for pytorch projects? Or is there other containers for that? How can I find them?

8

FYI, aarch64 should be the same as arm64; a.k.a., ARM 64-bit. Jetson hardware itself is likely somewhat different in some ways from other hardware (even two RPis are different from each other in most cases). Running on Linux I don’t think it matters if it is Jetson or not, it is still 64-bit Linux aarch64/arm64.

Unless an emulation is specific to some particular hardware I doubt any emulation is a perfect test platform. If you run a given application on the emulation, and it turns out the same as the actual hardware, then the emulation is probably a time saver and useful. Just test it on whatever you are working with.

9

Hi

The container is for cross-compiling.
This means you will compile a Jetson app with the container.
But this app cannot run on the desktop

For better development, you might want to use the same container that supports different platforms.

For example, design the app with nvcr.io/nvidia/pytorch:25.06-py3, which can run on a desktop.
And the same app is expected to be runnable on Jetson but with the nvcr.io/nvidia/pytorch:25.06-py3-igpu container.

Thanks.

11

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