Supercomputers/Systems/Mustang (AFRL)
USA · operational · hpc
Mustang (AFRL)
Also known as AFRL DSRC Mustang
Operated by AFRL DSRC at AFRL DSRC (Wright-Patterson AFB) .
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
- 3.22 PFlop/s
- Rpeak
- 4.78 PFlop/s
- Rmax ÷ Rpeak
- 67.4%
- Cores
- 55,296
- Power
- -
- Per watt
- - 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 | - HPE SGI 8600, four-system AFRL DSRC purchase (unclassified Mustang) | Hewlett Packard Enterprise | - | Reported | top500.orgnextplatform.com |
| Accelerator | NVIDIA Tesla P100 Accelerators CUDA (Pascal GP100) · 3,584C · 1.48 GHz · 16 nm · 300 W NVIDIA Tesla P100 GPUs in 24 nodes | NVIDIA | - | Reported | nextplatform.com |
| CPU | - Intel Xeon Platinum 8168 24C 2.7GHz, 1,176 dual-socket nodes | Intel | - | Reported | nextplatform.comtop500.org |
| Interconnect | Omni-Path Architecture Interconnect Fat tree · 100 Gb/s per port · About 1 microsecond Intel Omni-Path | Cornelis Networks | - | Reported | top500.orgnextplatform.com |
| Storage | Lustre Storage Parallel filesystem, separate metadata and object servers 8.4 PB Lustre storage | DDN | - | Reported | nextplatform.com |
| Operating system | Red Hat Enterprise Linux Operating systems Linux distribution · Open source, commercial support subscription Red Hat Enterprise Linux | Red Hat | - | Reported | top500.org |
The read
The unclassified HPE SGI 8600 at the Air Force Research Laboratory DoD Supercomputing Resource Center, bought alongside three classified sister systems (Voodoo, Shadow and Spectre) in the HPCMP's seven-system HPE contract valued at 57 million dollars including five years of support. The Next Platform reports 1,176 dual-socket Xeon Platinum 8168 nodes, 24 nodes with NVIDIA Tesla P100 GPUs, Intel Omni-Path, 8.4 PB of Lustre storage and about 4.9 PFlop/s peak, operational in December 2018. TOP500 lists 55,296 cores, Rmax 3.22 PFlop/s and Rpeak 4.78 PFlop/s, first at rank 48 on the June 2018 list and last in June 2025 at rank 349. Peak, core count and operational date differ slightly between the two sources; TOP500 figures are recorded. The name Mustang is also used by an unrelated 2011 Los Alamos cluster already in the register.
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
AFRL DSRC (Wright-Patterson AFB)
- Location
- Wright-Patterson AFB, OH, USA
Change history
- 2026-10-07 Added Mustang, the AFRL DSRC unclassified HPE SGI 8600
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 | Omni-Path Architecture |
| nextplatform.com | names system and part | NVIDIA Tesla P100, Omni-Path Architecture, Lustre |
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.