ASCEND 310P vs ASCEND 910
The cheapest ASCEND 310P ($0.22) costs 83% less per hour than the cheapest ASCEND 910 ($1.28).
Per unit of performance, the ASCEND 310P is 22% cheaper: $1.37 versus $1.76 per A100-equivalent hour.
The ASCEND 910 has 32 GB of memory against 24 GB on the ASCEND 310P.
| Measure | ASCEND 310P | ASCEND 910 |
|---|---|---|
| Lowest per GPU-hour | $0.22 | $1.28 |
| Median per GPU-hour | $1.05 | $1.28 |
| On-demand from | $1.60 | — |
| Spot from | — | — |
| Reserved from | $0.22 | $1.28 |
| Lowest per A100-equivalent hour | $1.37 | $1.76 |
| Live offers | 7 | 1 |
| Providers | 2 | 1 |
| Memory | 24 GB LPDDR4X | 32 GB HBM2 |
| Memory bandwidth | 205 GB/s | 1,228 GB/s |
| FP16 / BF16 | 70 TFLOPS | 256 TFLOPS |
| FP8 | — | — |
| Performance vs A100 80GB | 0.16× | 0.73× |
| Architecture | 达芬奇 (Da Vinci) / 昇腾 310P | 达芬奇 (Da Vinci) |
| Power | 72 W | 310 W |