TensorFlow For Jetson Platform
This document describes the key features, software enhancements and improvements, and known issues regarding Tensorflow 1.15.5 and 2.9.1 on the Jetson platform.
Compatibility
TensorFlow Version | NVIDIA TensorFlow Container | JetPack Version |
---|---|---|
2.9.1 | 22.06 | 5.0.1 |
2.8.0 | 22.05, 22.04, 22.03 | 5.0 |
2.7.0 | 22.01 | 4.6.1 |
2.6.2 | 21.12 | 4.6 |
2.6.0 | 21.11, 21.09 | 4.6 |
2.5.0 | 21.08, 21.07 | 4.6 |
21.06 | 4.5 | |
2.4.0 | 21.05, 21.04, 21.03, 21.02 | 4.5 |
2.3.1 | 20.12 | 4.5 |
20.12, 20.11, 20.10 | 4.4.x | |
2.3.0 | 20.09 | 4.4.x |
2.2.0 | 20.08, 20.07, 20.06 | 4.4.x |
2.1.0 | 20.04 | 4.4.x |
20.03, 20.02 | 4.3 | |
1.15.5 | 22.06 | 5.0.1 |
22.05, 22.04, 22.03 | 5.0 | |
22.01 | 4.6.1 | |
21.12, 21.11, 21.09, 21.08, 21.07 | 4.6 | |
21.06, 21.05, 21.04, 21.03, 21.02 | 4.5 | |
1.15.4 | 20.12 | 4.5 |
20.12, 20.11, 20.10 | 4.4.x | |
1.15.3 | 20.09, 20.08, 20.07 | 4.4.x |
1.15.2 | 20.06, 20.04 | 4.4.x |
20.03, 20.02 | 4.3 | |
Older packages below are installed as tensorflow-gpu; more recent releases above as tensorflow. | ||
2.0.0 | 20.01, 19.12 | 4.3 |
19.11 | 4.2.x | |
1.15.0 | 20.01, 19.12 | 4.3 |
19.11 | 4.2.x | |
1.14.0 | 19.10, 19.09, 19.07 | 4.2.x |
1.14.0 | 19.09 | 3.3.1 |
1.13.1 | 19.05, 19.04, 19.03 | 4.2.x |
-
If you are using TensorRT with TensorFlow, ensure you are familiar with the TensorRT Container Release Notes for any known issues.
Using TensorFlow With The Jetson Platform
- Memory
-
If you observe any out-of-memory problems in TensorFlow, you can use a custom configuration to limit the amount of memory TensorFlow tries to allocate. This can be accomplished by allowing the GPU memory allocation to grow, or setting a hard limit on the amount of memory the allocator will attempt to use. Depending on which version of the framework you are using, please see either the TensorFlow 2 guide or the archived TensorFlow 1.x documentation for details.
- Storage
-
If you need more storage, we recommend connecting an external SSD via SATA on AGX Orin or Xavier devices, or USB on the NX series.
- Operators
-
In TensorFlow 1.x, if you want to see which operators of your graph are placed on your module, use tf.ConfigProto(log_device_placement=True) to see all the device placements.