Parts/CPUs/AMD EPYC 7A53 (Trento)
Why this part
Trento is not a catalogue product. It is a Milan-generation part with the I/O die reworked to carry Infinity Fabric off the socket to the MI250X accelerators, giving Frontier cache-coherent CPU-to-GPU links from a single vendor. This is AMD building what IBM built with POWER9 and NVLink, on the other side of the market, and it is the reason Frontier's node has no PCIe bottleneck between host and device. Buying it means buying AMD at both layers: there is no configuration of this node with someone else's accelerator in it.
Systems using this part
- Systems
- 12
- Rmax underneath
- 2.32 EFlop/s
- Units recorded
- 9,600
2 of 12 edges disclose a quantity
| System | Role | Country | Units | System Rmax | Supplied by | Rank |
|---|---|---|---|---|---|---|
| HPC6 2024-11 | CPU | Italy | - | 477.9 PFlop/s | AMD | #8 → |
| LUMI 2022-06 | CPU | Finland | - | 379.7 PFlop/s | AMD | #3 → |
| Frontier 2022-05 | CPU | USA | 9,408 | 1.35 EFlop/s | AMD | #8 → |
| Adastra 2022 | CPU | France | - | 46.10 PFlop/s | AMD | #55 → |
| Setonix 2022 | CPU | Australia | - | 27.16 PFlop/s | AMD | #81 → |
| Dardel 2022 | CPU | Sweden | - | 9.90 PFlop/s | AMD | #77 → |
| Tioga 2022 | CPU | USA | - | 4.55 PFlop/s | AMD | #121 → |
| rzVernal 2022 | CPU | USA | - | - | AMD | - |
| Tenaya 2022 | CPU | USA | - | 3.41 PFlop/s | AMD | #200 → |
| Frontier TDS 2021 | CPU | USA | - | 19.20 PFlop/s | AMD | #1 → |
| Crusher | CPU | USA | 192 | - | AMD | - |
| Ruth | CPU | USA | - | - | AMD | - |
“Units recorded” sums only the edges where a public source states a quantity. It is a floor, not a total, and should never be read as installed base.