Flopper.io GPU Comparison Sheet
November 11, 2026
flopper.io/compare/huawei-ascend-910a-vs-huawei-atlas-350Huawei Ascend 910A vs Huawei Atlas 350
Side by side datacenter GPU specifications from the Flopper.io database.
Overview
| Huawei Ascend 910A | Huawei Atlas 350 | |
|---|---|---|
| FP32 TFLOPS | 16.0 | — |
| Memory | 32 GB | 112 GB |
| Memory Bandwidth | 1.2 TB/s | 1.4 TB/s |
| TDP | 310 W | 600 W |
Performance by Precision
| Precision | Huawei Ascend 910A peak TFLOPS (dense) | Huawei Atlas 350 peak TFLOPS (dense) |
|---|---|---|
| FP64 | — | — |
| FP32 | 16.0 | — |
| TF32 | — | — |
| BF16 | — | 425.0 |
| FP16 | 256.0 | 425.0 |
| FP8 | — | 804.0 |
| FP6 | — | 804.0 |
| FP4 | — | 1561.0 |
| INT8 | 512.0 | 804.0 |
Specifications
| Huawei Ascend 910A | Huawei Atlas 350 | |
|---|---|---|
| Architecture | Da Vinci | Da Vinci v3 |
| Process Node | TSMC 7nm | — |
| Launch Year | 2019 | 2026 |
| Form Factor | OAM | Custom |
| Memory Type | HBM2 | HiBL 1.0 |
| Interface Width | — | — |
| Interconnect | HCCS | Unified Bus 2.0 |
| TDP | 310 W | 600 W |
| FP32 Efficiency | 0.05 TFLOPS/W | — |