Of the two Jetson modules the NRU-222S supports — Jetson AGX Orin 32GB and Jetson AGX Orin 64GB — the Jetson AGX Orin 64GB has more CUDA cores, and it has 64 GB of memory (LPDDR5). neousys-te... +1neousys-tech.com ↗“Jetson AGX Orin 32GB, Jetson AGX Orin 64GB”nvidia.com ↗“2048 CUDA + 64 Tensor cores, 12-core Cortex-A78AE, 64GB LPDDR5”
Specifically:
| Module | CUDA cores | Memory | Memory bandwidth |
|---|---|---|---|
| Jetson AGX Orin 64GB | 2048 | 64 GB LPDDR5 | 204.8 GB/s |
| Jetson AGX Orin 32GB | 1792 | 32 GB LPDDR5 | 204.8 GB/s |
nvidia.com +1nvidia.com ↗“2048 CUDA + 64 Tensor cores, 12-core Cortex-A78AE, 64GB LPDDR5”nvidia.com ↗“1792 CUDA + 56 Tensor cores, 8-core Cortex-A78AE, 32GB LPDDR5”
Both are Ampere-architecture modules with dual NVDLA engines. The 64GB module also delivers higher AI throughput (275 INT8 TOPS sparse vs. 200 INT8 TOPS sparse) and a higher FP32 rate (5.3 TFLOPS vs. 3.8 TFLOPS). nvidia.com +3nvidia.com ↗“Ampere SoM”nvidia.com ↗“dual NVDLA”nvidia.com ↗“275 INT8 TOPS (sparse)”nvidia.com ↗“200 INT8 TOPS (sparse)”