L40 vs RTX 5090
The cheapest RTX 5090 ($0.22) costs 34% less per hour than the cheapest L40 ($0.33).
Per unit of performance, the RTX 5090 is 55% cheaper: $0.24 versus $0.52 per A100-equivalent hour.
The L40 has 48 GB of memory against 32 GB on the RTX 5090.
| Measure | NVIDIA L40 48GB | NVIDIA GeForce RTX 5090 |
|---|---|---|
| Lowest per GPU-hour | $0.33 | $0.22 |
| Median per GPU-hour | $0.99 | $0.87 |
| On-demand from | $0.45 | $0.34 |
| Spot from | $0.33 | $0.25 |
| Reserved from | $0.56 | $0.22 |
| Lowest per A100-equivalent hour | $0.52 | $0.24 |
| Live offers | 111 | 405 |
| Providers | 32 | 84 |
| Memory | 48 GB GDDR6 | 32 GB GDDR7 |
| Memory bandwidth | 864 GB/s | 1,792 GB/s |
| FP16 / BF16 | 181 TFLOPS | 209.5 TFLOPS |
| FP8 | 362 TFLOPS | 419 TFLOPS |
| Performance vs A100 80GB | 0.63× | 0.92× |
| Architecture | Ada Lovelace | Blackwell |
| Power | 300 W | 575 W |