NVIDIA Jetson Orin Series
Overview
The NVIDIA Jetson Orin is a family of SoMs (System-on-Module) designed for edge AI and robotics applications. Based on the Ampere architecture GPU + ARM Cortex-A78AE CPU, it delivers AI compute ranging from 67 TOPS to 275 TOPS. The Orin series includes the Orin Nano, Orin NX, and AGX Orin sub-families.
Orin Nano Series
| Spec | Orin Nano 4GB | Orin Nano 8GB |
|---|
| AI Compute | 34 TOPS (INT8) | 67 TOPS (INT8) |
| GPU | 512 CUDA Cores + 16 Tensor Cores | 1,024 CUDA Cores + 32 Tensor Cores |
| CPU | 6× Arm Cortex-A78AE 1.5 MB L2 + 4 MB L3 | 6× Arm Cortex-A78AE 1.5 MB L2 + 4 MB L3 |
| Memory | 4 GB LPDDR5 (51 GB/s) | 8 GB LPDDR5 (102 GB/s) |
| Power | 7-25 W | 7-25 W |
| Price | $199 | $249 (Super Developer Kit) |
Orin NX Series
| Spec | Orin NX 8GB | Orin NX 16GB |
|---|
| AI Compute | 117 TOPS | 157 TOPS |
| GPU | 1,792 CUDA + 56 Tensor Cores | 2,048 CUDA + 64 Tensor Cores |
| CPU | 8× Arm Cortex-A78AE 2.0 GHz | 8× Arm Cortex-A78AE 2.0 GHz |
| DLA | 1× NVDLA v2 1.23 GHz | 2× NVDLA v2 1.4 GHz |
| PVA | 1× PVA v2 | 1× PVA v2 |
| Memory | 8 GB LPDDR5 (102 GB/s) | 16 GB LPDDR5 (102 GB/s) |
| Power | 10-25 W | 10-25 W |
AGX Orin Series
| Spec | AGX Orin 32GB | AGX Orin 64GB |
|---|
| AI Compute | 200 TOPS | 275 TOPS |
| GPU | 2,048 CUDA + 64 Tensor Cores | 2,048 CUDA + 64 Tensor Cores |
| CPU | 12× Arm Cortex-A78AE 2.2 GHz | 12× Arm Cortex-A78AE 2.2 GHz |
| DLA | 2× NVDLA v2 1.6 GHz | 2× NVDLA v2 1.6 GHz |
| PVA | 1× PVA v2 | 1× PVA v2 |
| Memory | 32 GB LPDDR5 (204.8 GB/s) | 64 GB LPDDR5 (204.8 GB/s) |
| Power | 15-60 W | 15-60 W |
Key Features
- Full CUDA ecosystem: Same software stack as server GPUs
- Multimodal sensing: ISP + PVA (vision accelerator) + DLA (deep learning accelerator)
- Jetson Orin Nano Super: Software upgrade boosts 40 TOPS to 67 TOPS
- Long-life support: 10+ year supply commitment
Use Cases
- Robotics (AMR, humanoid robots)
- Industrial vision and inspection
- Autonomous driving (edge)
- Smart cities (video analytics)
- Drones