L40S vs V100 32GB
The cheapest V100 32GB ($0.017) costs 96% less per hour than the cheapest L40S ($0.38).
Per unit of performance, the V100 32GB is 90% cheaper: $0.04 versus $0.40 per A100-equivalent hour.
The L40S has 48 GB of memory against 32 GB on the V100 32GB.
| Measure | NVIDIA L40S 48GB | NVIDIA Tesla V100 32GB |
|---|---|---|
| Lowest per GPU-hour | $0.38 | $0.017 |
| Median per GPU-hour | $1.30 | $2.32 |
| On-demand from | $0.38 | $0.017 |
| Spot from | $0.45 | $0.51 |
| Reserved from | $0.49 | $0.19 |
| Lowest per A100-equivalent hour | $0.40 | $0.04 |
| Live offers | 284 | 193 |
| Providers | 85 | 43 |
| Memory | 48 GB GDDR6 | 32 GB HBM2 |
| Memory bandwidth | 864 GB/s | 900 GB/s |
| FP16 / BF16 | 362 TFLOPS | 125 TFLOPS |
| FP8 | 733 TFLOPS | — |
| Performance vs A100 80GB | 0.96× | 0.42× |
| Architecture | Ada Lovelace | Volta |
| Power | 350 W | 300 W |