NVIDIA Jetson AGX Thor - Performance Benchmarking
Follow Us On
 | The NVIDIA Jetson AGX Thor documentation from RidgeRun is presently being developed. |
| Sample | CPU | GPU |
|---|---|---|
| One frame deepstream reference app | 14.28% | 37.92% |
| Four frame deepstream reference app | 13.35% | 51.41% |
| Thirty frame deepstream reference app | 15.86% | 82.69% |
| License Plate detection and recognition | 2.89% | 10.13% |
| Parallel multiple models | 6.82% | 0% |
| Text recognition with OCD/OCR models | 1.35% | 85.13% |
| Object embedding vector generation peoplenet | 10.26% | 86.73% |
| Object embedding vector generation | 2.58% | 92.24% |
| Pose classification | 6.8% | 0% |