A30 vs B200
The cheapest A30 ($0.15) costs 85% less per hour than the cheapest B200 ($0.99).
Per unit of performance, the B200 is 53% cheaper: $0.14 versus $0.30 per A100-equivalent hour.
The B200 has 180 GB of memory against 24 GB on the A30.
| Measure | NVIDIA A30 24GB | NVIDIA B200 180GB SXM |
|---|---|---|
| Lowest per GPU-hour | $0.15 | $0.99 |
| Median per GPU-hour | $0.64 | $7.89 |
| On-demand from | $0.22 | $0.99 |
| Spot from | $0.15 | $1.16 |
| Reserved from | $0.23 | $3.89 |
| Lowest per A100-equivalent hour | $0.30 | $0.14 |
| Live offers | 32 | 146 |
| Providers | 17 | 63 |
| Memory | 24 GB HBM2 | 180 GB HBM3e |
| Memory bandwidth | 933 GB/s | 8,000 GB/s |
| FP16 / BF16 | 165 TFLOPS | 2,250 TFLOPS |
| FP8 | — | 4,500 TFLOPS |
| Performance vs A100 80GB | 0.5× | 6.96× |
| Architecture | Ampere | Blackwell |
| Power | 165 W | 1,000 W |