INFERENTIA 2 vs RTX 5090
| Measure | AWS Inferentia2 | NVIDIA GeForce RTX 5090 |
|---|---|---|
| Lowest per GPU-hour | — | $0.22 |
| Median per GPU-hour | — | $0.87 |
| On-demand from | — | $0.34 |
| Spot from | — | $0.25 |
| Reserved from | — | $0.22 |
| Lowest per A100-equivalent hour | — | $0.24 |
| Live offers | 0 | 405 |
| Providers | 0 | 84 |
| Memory | 32 GB HBM2e | 32 GB GDDR7 |
| Memory bandwidth | 820 GB/s | 1,792 GB/s |
| FP16 / BF16 | 190 TFLOPS | 209.5 TFLOPS |
| FP8 | 190 TFLOPS | 419 TFLOPS |
| Performance vs A100 80GB | 0.52× | 0.92× |
| Architecture | Inferentia2 (NeuronCore v2) | Blackwell |
| Power | — | 575 W |
Where the offers sit
No live offers right now.