Add support for JetPack 6.2 build #3453
Conversation
github-actions
bot
added
documentation
| --------------------- | ||
| Nvida JetPack 6.1 is the latest production release ofJetPack 6. | ||
| Nvida JetPack 6.2 is the latest production release ofJetPack 6. | ||
| With this release it incorporates: | ||
| CUDA 12.6 | ||
| TensorRT 10.3 |
There was a problem hiding this comment.
@zewenli98 Do you know of any incompatibilities between 10.3 and 10.9 (minus new APIs)?
There was a problem hiding this comment.
6.1 is the same as 6.2.
6.2 is an update for super nano
There was a problem hiding this comment.
Besides add_unsqueeze was added in TRT 10.7, I don't remember any incompatibilities. (add_unsqueeze is being used in the converter)
There was a problem hiding this comment.
@johnnynunez Thanks for opening this, this is very helpful. We will test it internally and provide additional feedback
| cmd.append("--platforms=//toolchains:jetpack_6.1") | ||
| print("Jetpack version: 6.1") | ||
| elif JETPACK_VERSION == "6.2": | ||
| cmd.append("--platforms=//toolchains:jetpack_6.2") |
There was a problem hiding this comment.
We might want to bind version 6.1 and 6.2 to like a 6.x bazel toolchain if there arent really any distinguishing dependencies.
There was a problem hiding this comment.
I changed it because 6.1 is the same as 6.2
6.2 is the “super nano” update
| @@ -13,6 +13,6 @@ constraint_value( | |||
| ) | |||
|
|
|||
| constraint_value( | |||
| name = "6.1", | |||
| name = "6.2", | |||
Add support for the jetpack6.2 build.
Currently jetpack 6.2 has:
cuda: 12.6
python: 3.10
tensorrt: 10.3
DLFW: 24.08 (pytorch: 2.6.0)
Jetson now distribute wheels on https://pypi.jetson-ai-lab.dev/
jetpack6.2 wheels: https://pypi.jetson-ai-lab.dev/jp6/cu126