ASCEND 310P vs RTX 4090
The cheapest RTX 4090 ($0.16) costs 28% less per hour than the cheapest ASCEND 310P ($0.22).
Per unit of performance, the RTX 4090 is 82% cheaper: $0.25 versus $1.37 per A100-equivalent hour.
| Measure | ASCEND 310P | NVIDIA GeForce RTX 4090 |
|---|---|---|
| Lowest per GPU-hour | $0.22 | $0.16 |
| Median per GPU-hour | $1.05 | $0.58 |
| On-demand from | $1.60 | $0.19 |
| Spot from | — | $0.16 |
| Reserved from | $0.22 | $0.18 |
| Lowest per A100-equivalent hour | $1.37 | $0.25 |
| Live offers | 7 | 427 |
| Providers | 2 | 91 |
| Memory | 24 GB LPDDR4X | 24 GB GDDR6X |
| Memory bandwidth | 205 GB/s | 1,008 GB/s |
| FP16 / BF16 | 70 TFLOPS | 165.2 TFLOPS |
| FP8 | — | 330.3 TFLOPS |
| Performance vs A100 80GB | 0.16× | 0.63× |
| Architecture | 达芬奇 (Da Vinci) / 昇腾 310P | Ada Lovelace |
| Power | 72 W | 450 W |