Parts/Accelerators/Intel Data Center GPU Max 1550
Accelerators · Ponte Vecchio · 2023
Intel Data Center GPU Max 1550
128 GB HBM2e across 2 stacks and 47 tiles. Six per node in the Aurora design.
Why this part
Ponte Vecchio is the most complex processor anyone has shipped: 47 tiles across five process nodes from two different foundries, including Intel's own and TSMC's. Aurora is built on 63,744 of them, and the machine arrived years late in large part because this part did. It is the strongest available illustration that a supercomputer procurement is a bet on a silicon roadmap, and that when the roadmap slips there is no substitute to switch to.
Systems using this part
- Systems
- 5
- Rmax underneath
- 1.05 EFlop/s
- Units recorded
- 66,024
| System | Role | Country | Units | System Rmax | Supplied by | Rank |
|---|---|---|---|---|---|---|
| Dawn 2024-01 | Accelerator | United Kingdom | 1,024 | 19.46 PFlop/s | Intel | #40 → |
| Stampede3 2024-01 | Accelerator | USA | - | - | Intel | - |
| Aurora 2023-12 | Accelerator | USA | 63,744 | 1.01 EFlop/s | Intel | #2 → |
| Clementina XXI | Accelerator | Argentina | 296 | 5.39 PFlop/s | Intel | #199 → |
| SuperMUC-NG Phase 2 | Accelerator | Germany | 960 | 17.19 PFlop/s | Intel | #51 → |
“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.
Related news
- Aurora enters service at Argonne · Argonne Leadership Computing FacilityEight years after contract award, following the cancellation of its original Xeon Phi design and a redesign around Ponte Vecchio. The longest announcement-to-first-light gap in this dataset.