Hello, I have a question about the jetson device. Public information shows that the jetpack version limits the upper limit of the pytorch version that can be installed on the orin host. If cuda and cudnn are installed on the host, and jetpack has some compatibility requirements for cuda and cudnn, the versions do need to be unified. But if I install docker on jetson, and install cuda, cudnn and pytorch in docker, are their versions still restricted by the jetpack version? I am very confused about the relationship between them and hope to get a detailed answer. The real scenario is that the jetpack version on my orin is 5.1.3, which stipulates that the upper limit of the pytorch version is 2.0. If I want to use the 2.3 version of the pytorch model, can I only upgrade the jetpack version?
Refer this link for more information:
This means that if I want to use a higher version of pytorch, I can only upgrade the jetpack version, regardless of whether the model is running on the host machine or in the docker container, right?
Hi,
You can find PyTorch 2.2 for the JetPack 5 environment below:
For a newer release, you will need to either build it from the source or upgrade the Jetpack to a newer release.
Thanks.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.