Parts/CPUs/AMD EPYC 7742 (Rome)
Why this part
The flagship Rome part, and the one that shows how differently the same chip gets used. ARCHER2 and Hawk run it as the entire machine, with no accelerator at all, for a user base whose codes are memory-bandwidth bound rather than arithmetic bound. NVIDIA's Selene runs the same part purely as a host under A100s. Nothing in a spec string distinguishes those two roles, which is the gap this dataset exists to fill.
Systems using this part
- Systems
- 36
- Rmax underneath
- 295.5 PFlop/s
- Units recorded
- 4,433
6 of 36 edges disclose a quantity
| System | Role | Country | Units | System Rmax | Supplied by | Rank |
|---|---|---|---|---|---|---|
| HEMUS 2023-10 | CPU | Bulgaria | - | 2.53 PFlop/s | AMD | #359 → |
| AIRAWAT-PSAI 2023-05 | CPU | India | 1,271 | 8.50 PFlop/s | AMD | #75 → |
| Sejong 2023 | CPU | South Korea | - | 32.97 PFlop/s | AMD | #22 → |
| ECMWF HPCF-AA 2022-10 | CPU | Italy | - | 6.67 PFlop/s | AMD | #117 → |
| KT DGX SuperPOD 2022 | CPU | South Korea | - | 10.38 PFlop/s | AMD | #58 → |
| Dardel 2022 | CPU | Sweden | - | 9.90 PFlop/s | AMD | #77 → |
| ARCHER2 2021-11 | CPU | United Kingdom | - | 19.54 PFlop/s | AMD | #22 → |
| Sulis 2021-11 | CPU | United Kingdom | - | - | AMD | - |
| Tethys 2021-11 | CPU | USA | - | 2.26 PFlop/s | AMD | #3 → |
| Cambridge-1 2021-07-07 | CPU | United Kingdom | - | 9.68 PFlop/s | AMD | #41 → |
| BioHive-1 2021-06 | CPU | USA | - | 4.90 PFlop/s | AMD | #84 → |
| Bridges-2 2021-03 | CPU | USA | - | - | AMD | - |
| Earth Simulator 4 2021-03 | CPU | Japan | 2,140 | 9.99 PFlop/s | AMD | #39 → |
| Athena 2021 | CPU | Poland | 96 | 5.05 PFlop/s | AMD | #105 → |
| Cactus 2021 | CPU | USA | - | 10.01 PFlop/s | AMD | #37 → |
| Berzelius 2021 | CPU | Sweden | - | 5.25 PFlop/s | AMD | #82 → |
| MTS GROM 2021 | CPU | Russia | - | 2.26 PFlop/s | AMD | #240 → |
| Expanse 2020-12 | CPU | USA | - | 2.48 PFlop/s | AMD | #153 → |
| Betzy 2020-11 | CPU | Norway | - | 4.72 PFlop/s | AMD | #55 → |
| PARAM Siddhi-AI 2020-11 | CPU | India | - | 4.62 PFlop/s | AMD | #62 → |
| Selene 2020-06 | CPU | USA | - | 63.46 PFlop/s | AMD | #36 → |
| Hawk 2020-02 | CPU | Germany | - | 19.33 PFlop/s | AMD | #16 → |
| Big Red 200 2020-01 | CPU | USA | - | 2.43 PFlop/s | AMD | #205 → |
| JURECA-DC 2020 | CPU | Germany | - | 9.33 PFlop/s | AMD | #45 → |
| Dogwood 2020 | CPU | USA | - | 10.01 PFlop/s | AMD | #38 → |
| Taranis 2020 | CPU | France | - | 8.19 PFlop/s | AMD | #30 → |
| Belenos 2020 | CPU | France | - | 7.68 PFlop/s | AMD | #29 → |
| Tinkercliffs 2020 | CPU | USA | 14 | - | AMD | - |
| NVIDIA DGX SuperPOD (A100) 2020 | CPU | USA | - | 2.36 PFlop/s | AMD | #170 → |
| Aitken 2019 | CPU | USA | - | 9.07 PFlop/s | AMD | #58 → |
| Euler 2014 | CPU | Switzerland | 216 | - | AMD | - |
| Christofari Neo | CPU | Russia | - | 11.95 PFlop/s | AMD | #43 → |
| ALTAY | CPU | Turkey | - | - | AMD | - |
| Imperial CX3 | CPU | United Kingdom | 696 | - | AMD | - |
| Technion Zeus | CPU | Israel | - | - | AMD | - |
| UBELIX | CPU | Switzerland | - | - | 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.