H100 SXM 80GB vs RTX 4000 ADA 20GB
The cheapest RTX 4000 ADA 20GB ($0.067) costs 87% less per hour than the cheapest H100 SXM 80GB ($0.50).
Per unit of performance, the H100 SXM 80GB is 31% cheaper: $0.17 versus $0.24 per A100-equivalent hour.
The H100 SXM 80GB has 80 GB of memory against 20 GB on the RTX 4000 ADA 20GB.
| Measure | NVIDIA H100 80GB SXM5 | NVIDIA RTX 4000 Ada Generation 20GB |
|---|---|---|
| Lowest per GPU-hour | $0.50 | $0.067 |
| Median per GPU-hour | $3.51 | $0.47 |
| On-demand from | $0.58 | $0.067 |
| Spot from | $0.50 | — |
| Reserved from | $1.41 | $0.18 |
| Lowest per A100-equivalent hour | $0.17 | $0.24 |
| Live offers | 317 | 61 |
| Providers | 83 | 24 |
| Memory | 80 GB HBM3 | 20 GB GDDR6 |
| Memory bandwidth | 3,350 GB/s | 360 GB/s |
| FP16 / BF16 | 989 TFLOPS | 81.9 TFLOPS |
| FP8 | 1,979 TFLOPS | 163.8 TFLOPS |
| Performance vs A100 80GB | 3× | 0.28× |
| Architecture | Hopper | Ada Lovelace |
| Power | 700 W | 130 W |