Supercomputers/Systems/Tursa
United Kingdom · operational · hpc
Tursa
Also known as DiRAC Tursa; Extreme Scaling GPU
Operated by EPCC at Advanced Computing Facility, Edinburgh .
Phase 1 seed dataset, compiled by hand. These rows were built from public operator, laboratory and vendor sources. A mechanical second-reader pass has since fetched every cited source: 1092 of 1153 systems have a readable citation that names them, and 40 are genuinely weakly sourced. Every claim carries its source and a confidence tier. Treat anything below verified as a lead, not a citation.
Measured performance
- Rmax
- 5.23 PFlop/s
- Rpeak
- 8.58 PFlop/s
- Rmax ÷ Rpeak
- 61.0%
- Cores
- 55,552
- Power
- 243.95 kW
- Per watt
- 21.4 GF/W
Figures are as last publicly reported for the configuration described below, not a live measurement. Where a system was upgraded in place, the post-upgrade configuration is shown and the earlier one appears in the timeline.
What this machine is made of
One row per supplier relationship. “Supplier at build” is the company that shipped the part at the time; where that company has since been acquired, the parent it rolls up to today is shown beside it. That distinction is what makes ticker-level aggregation possible across a thirty-year dataset.
| Role | Part | Supplier at build | Quantity | Confidence | Source |
|---|---|---|---|---|---|
| Integrator | - Eviden BullSequana XH2000 (TOP500 manufacturer Bull) | Eviden | - | Verified | top500.orgepcc.ed.ac.uk |
| Accelerator | NVIDIA A100 SXM4 40GB Accelerators CUDA (Ampere GA100) · 6,912C · 1.41 GHz · 7 nm · 400 W NVIDIA A100 40GB (114 nodes x 4); 64 further nodes carry A100 80GB | NVIDIA | - | Verified | top500.orgepcc.ed.ac.uk |
| CPU | - AMD EPYC Rome host CPUs (TOP500 7302 32C 3GHz; EPCC 7272 GPU nodes, 7H12 CPU nodes) | AMD | - | Reported | top500.orgepcc.ed.ac.uk |
| Interconnect | NVIDIA InfiniBand HDR Interconnect Fat tree or dragonfly+ · 200 Gb/s per port · About 0.6 microseconds 4x 200 Gbps NVIDIA InfiniBand HDR per GPU node, fat tree | NVIDIA | - | Verified | top500.orgepcc.ed.ac.uk |
| Cooling | Direct liquid cooling, warm water Cooling Direct-to-chip cold plates, facility water loop · Removes 90+ percent of node heat to liquid Water-cooled BullSequana XH2000 racks | Eviden | - | Reported | epcc.ed.ac.uk |
| Operating system | Red Hat Enterprise Linux Operating systems Linux distribution · Open source, commercial support subscription Red Hat Enterprise Linux | Red Hat | - | Reported | epcc.ed.ac.uk |
The read
The University of Edinburgh's DiRAC Extreme Scaling GPU system, hosted by EPCC at its Advanced Computing Facility: five water-cooled Eviden BullSequana XH2000 racks holding 184 GPU nodes (114 with four A100-40, 64 with four A100-80) and 6 CPU nodes, four 200 Gbps InfiniBand interfaces per GPU node in a fat tree, 8 PB of DDN Lustre and RHEL. TOP500 lists 55,552 cores of AMD EPYC 7302 with A100 40GB, HDR InfiniBand, Rmax 5.23 PFlop/s, Rpeak 8.58 PFlop/s and 244 kW, installed 2021; EPCC names the GPU-node CPU as EPYC 7272, which conflicts, so the CPU part is left generic. It debuted at 95th on the November 2021 TOP500, its best rank, and was 296th in June 2026.
Timeline
Public list appearances
Pointers only: rank and edition, linking to the canonical entry. We do not reproduce list tables. See sourcing policy.
Site & facility
Advanced Computing Facility, Edinburgh
- Location
- Edinburgh, United Kingdom
- Cooling
- Direct liquid cooling
- Commissioned
- 1990
Change history
- 2026-10-02 Added Tursa, the DiRAC Extreme Scaling GPU system at EPCC, 95th on the November 2021 TOP500
Source check
We fetched this system's own citations and recorded whether each page actually mentions it. This is a corroboration signal, not a fact check, and it is published so you can see how well the sourcing holds up rather than take it on trust.
| Cited source | Result | Found on the page |
|---|---|---|
| top500.org | names system and part | NVIDIA A100 SXM4 40GB, NVIDIA InfiniBand HDR |
| epcc.ed.ac.uk | names system and part | NVIDIA A100 SXM4 40GB, NVIDIA InfiniBand HDR, Direct liquid cooling, warm water |
Checked 2026-10-09 by pnpm verify. Re-run it and the table changes with the web.
Further reading
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.
Inside a node
Latency-optimised versus throughput-optimised processors, NUMA, and the end of the host-to-device copy.
The interconnect
Topologies, why latency and tail behaviour matter more than bandwidth, and how the fabric market consolidated.