RTX 5090 vs V100S 32GB
The cheapest V100S 32GB ($0.15) costs 33% less per hour than the cheapest RTX 5090 ($0.22).
Per unit of performance, the RTX 5090 is 25% cheaper: $0.24 versus $0.31 per A100-equivalent hour.
| Measure | NVIDIA GeForce RTX 5090 | NVIDIA Tesla V100S 32GB PCIe |
|---|---|---|
| Lowest per GPU-hour | $0.22 | $0.15 |
| Median per GPU-hour | $0.87 | $0.21 |
| On-demand from | $0.34 | $0.21 |
| Spot from | $0.25 | $0.15 |
| Reserved from | $0.22 | $0.17 |
| Lowest per A100-equivalent hour | $0.24 | $0.31 |
| Live offers | 405 | 25 |
| Providers | 84 | 4 |
| Memory | 32 GB GDDR7 | 32 GB HBM2 |
| Memory bandwidth | 1,792 GB/s | 1,134 GB/s |
| FP16 / BF16 | 209.5 TFLOPS | 130 TFLOPS |
| FP8 | 419 TFLOPS | — |
| Performance vs A100 80GB | 0.92× | 0.47× |
| Architecture | Blackwell | Volta |
| Power | 575 W | 250 W |