The Jetson Orin Nano 8GB has twice the 4GB module's figures in all three areas: 32 vs 16 Tensor Cores, 102 vs 51 GB/s Super-mode memory bandwidth, and 67 vs 34 Super-mode AI TOPS.
Jetson Orin Nano 4GB vs 8GB nvidia.com +1nvidia.com ↗“memory.memory_bandwidth_gb_per_s_super=51”nvidia.com ↗“memory.memory_bandwidth_gb_per_s_super=102”
| Spec | Jetson Orin Nano 4GB | Jetson Orin Nano 8GB |
|---|---|---|
| Tensor Cores | 16 | 32 |
| Memory bandwidth (Super mode) | 51 GB/s | 102 GB/s |
| AI performance (Super mode) | 34 TOPS | 67 TOPS |
Notes:
- The AI TOPS figures are INT8. For Orin products, they're sparse INT8 numbers.
- Both modules have Super-mode ratings because Super mode is available on the Jetson Orin Nano.
- Both are listed as Production Modules and use Ampere GPUs. nvidia.com +1nvidia.com ↗“compute.gpu_architecture=Ampere”nvidia.com ↗“compute.gpu_architecture=Ampere”