Supercomputers/Systems/Frontier
USA · upgraded · mixed
Frontier
Deep dive: Frontier: the first exaflop and its price
Also known as OLCF-5
Operated by Oak Ridge National Laboratory at Oak Ridge Leadership Computing Facility .
Researched with AI assistance, checked by scripts. These rows were researched by AI agents reading public operator, laboratory and vendor sources, appended by scripts and checked by automated validation; no subject-matter expert has reviewed them (how this site is made). A mechanical second-reader pass has since fetched every cited source: 1350 of 1581 systems have a readable citation that names them, and 64 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
- 1.35 EFlop/s
- Rpeak
- 2.06 EFlop/s
- Rmax ÷ Rpeak
- 65.8%
- Cores
- 9,066,176
- Power
- 24.61 MW
- Per watt
- 55.0 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.
How it compares
Ranked against the other machines in this dataset that report the same figure: those that first came into service within two years of Frontier, and those in USA. Computed from our tables; a metric is ranked only when at least eight peers report it.
| Metric | Value | Against its era | Against its country |
|---|---|---|---|
| Rmax | 1.35 EFlop/s | #2 of 246 · top 1% | #2 of 332 · top 1% |
| Energy efficiency | 55.0 GF/W | #20 of 130 · top 15% | #12 of 162 · top 7% |
| Rmax ÷ Rpeak | 65.8 % | #161 of 243 · bottom 35% | #241 of 331 · bottom 28% |
| Power draw | 24,607 kW | #4 of 133 · top 2% | #3 of 171 · top 1% |
Every metric is ranked largest-first, so a high power-draw rank means a big power draw, not a good one. Rmax ÷ Rpeak uses HPL-measured results only. Rankings move as machines are added; they describe this dataset, not the world.
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 Cray EX235a | Hewlett Packard Enterprise | - | Verified | olcf.ornl.govhpe.com |
| Accelerator | AMD Instinct MI250X Accelerators ROCm (CDNA 2) · 1.7 GHz · TSMC N6 · 560 W AMD Instinct MI250X, 4 per node | AMD | 37,632 reported | Verified | olcf.ornl.govhpe.com |
| CPU | AMD EPYC 7A53 (Trento) CPUs x86-64 · 64C · 2 GHz · 7 nm · 280 W AMD Optimized 3rd Generation EPYC 64C 2GHz | AMD | 9,408 reported | Reported | olcf.ornl.govhpe.com |
| Interconnect | HPE Slingshot-11 Interconnect Dragonfly · 200 Gb/s per port · Around 2 microseconds Slingshot-11 | Hewlett Packard Enterprise | - | Verified | olcf.ornl.govhpe.com |
| Storage | HPE Cray ClusterStor E1000 Storage Lustre appliance, flash and disk tiers · 700 PB on Frontier (Orion) · About 10 TB/s peak Orion, Lustre on ClusterStor E1000 | Hewlett Packard Enterprise | - | Reported | olcf.ornl.govhpe.com |
| Cooling | Direct liquid cooling, warm water Cooling Direct-to-chip cold plates, facility water loop · Removes 90+ percent of node heat to liquid 100% direct liquid cooled cabinet | Hewlett Packard Enterprise | - | Reported | olcf.ornl.govhpe.com |
| Operating system | HPE Cray Operating System Operating systems SUSE-derived Linux, container-based management · Proprietary layer over open-source Linux HPE Cray Operating System | Hewlett Packard Enterprise | - | Reported | olcf.ornl.govhpe.com |
Engineering, funding and access
What Frontier runs on, who paid for it and how people get time on it. Each line carries who said it and when.
Hardware
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Compute nodes | 9,856 nodes77 cabinets of 128 nodes each; the user guide states 9,856 AMD compute nodes | 2025 | Operator stated | docs.olcf.ornl.gov |
| Storage capacity | 679 PBOrion usable namespace; each compute node also has two 1.92 TB NVMe devices | 2025 | Operator stated | docs.olcf.ornl.gov |
| Interconnect and topology | HPE Slingshot-11, four 200 Gb/s NICs per node (800 Gb/s injection bandwidth) | 2025 | Operator stated | docs.olcf.ornl.govtop500.org |
Software
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Parallel filesystem | Orion Lustre (HPE ClusterStor), 679 PB usable namespace, plus node-local NVMe | 2025 | Operator stated | docs.olcf.ornl.gov |
| Job scheduler | Slurm | 2025 | Operator stated | docs.olcf.ornl.gov |
| Programming and software stack | HPE Cray OS 2.4 based on SUSE Linux Enterprise Server 15.4TOP500 lists the OS as HPE Cray OS | 2025 | Operator stated | docs.olcf.ornl.gov |
| Programming and software stack | HIP, OpenMP GPU offload, OpenACC and SYCL programming models with Lmod environment modules | 2025 | Operator stated | docs.olcf.ornl.gov |
| Programming and software stack | HPL run used gcc 12.3.0 and hipcc, ROCm 5.7.1, Intel MKL 2019 and cray-mpich 8.1.30 | 2024-11 | Operator stated | top500.org |
Funding
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Funded by | U.S. Department of Energy (Office of Science)Hosted by ORNL in the Oak Ridge Leadership Computing Facility | 2019-05 | Operator stated | ornl.gov |
| Stated cost | 600 USD mContract valued at more than $600 million for system and technology development, awarded to Cray | 2019-05 | Operator stated | ornl.gov |
| Procurement | First CORAL-2 award: contract with Cray announced 7 May 2019 covering technology development, a center of excellence, early-delivery systems, the main system and multi-year supportOriginally targeted delivery in 2021 and acceptance in 2022 | 2019-05-07 | Operator stated | ornl.gov |
Access
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| How researchers get time | INCITE, ALCC and Director's Discretionary allocations | 2025 | Operator stated | olcf.ornl.gov |
| Access page | www.olcf.ornl.gov/for-users/getting-started | 2025 | Operator stated | olcf.ornl.gov |
AI datacenter metrics
Figures beyond an accelerator count, each with who stated it. What can and cannot be compared across AI datacenters.
| Metric | Value | As of | Basis | Note | Source |
|---|---|---|---|---|---|
| Accelerators | 37,632 accelerators | 2024-02 | Operator stated | OLCF slides: 9,408 nodes x 4 MI250X modules (product computed). Each module has 2 GCDs, so the OS sees 75,264 GPUs. | olcf.ornl.gov |
| Stated investment | 0.6 USD bn | 2019-05 | Operator stated | DOE: total contract award valued at more than $600M for the system and technology development. Contract value, excludes facility upgrades. | energy.gov |
| Facility power | 40 MW | 2020-09 | Design target | OLCF: about 40 MW of power and cooling provided to the Frontier data centre (new 2.5-mile line). Facility build target, not a metered draw. | olcf.ornl.govolcf.ornl.gov |
| IT power | 30 MW | 2020-12 | Design target | OLCF program director before delivery: at peak Frontier will consume about 30 MW. Design-time estimate for the system, not metered. | olcf.ornl.gov |
Published benchmark results
- HPL-MxP 2025-11: HPL-MxP mixed-precision (rank 3), 11.39 EFlop/s (independent, 2025-11) hpl-mxp.orgtop500.org
- HPL-MxP 2026-06: HPL-MxP mixed-precision (rank 3), 11.4 EFlop/s (independent, 2026-06) top500.orgtop500.org
The read
The first system to clear an exaflop on HPL. Every part of the compute node comes from AMD (a custom Trento EPYC linked to four MI250X modules by Infinity Fabric rather than PCIe) with HPE supplying the cabinet, the Slingshot-11 fabric and the Orion storage system. Two suppliers account for essentially the entire machine. Figures shown are the post-upgrade configuration.
Timeline
- 2019 Announced source
- 2019-05-07 DOE announces a contract valued at more than $600 million with Cray (CORAL-2) to deliver Frontier to ORNL source
- 2022-05 First operational source
- 2022-06 List debut at #1 (June 2022) source
- 2024-11 Reported at 1.353 EFlop/s after node and configuration changes source
Public list appearances
Pointers only: rank and edition, linking to the canonical entry. We do not reproduce list tables. See sourcing policy.
Site & facility
Oak Ridge Leadership Computing Facility
- Location
- Oak Ridge, TN, USA
- Utility
- Tennessee Valley Authority
- Cooling
- Direct liquid cooling (warm water)
- Commissioned
- 1993
Change history
- 2026-10-04 AI benchmark record: hpl-mxp 2025-11
- 2026-10-04 AI benchmark record: hpl-mxp 2026-06
- 2026-09-22 Gained a Green500 rank pointer (#8 on the 2023-11 list)
Nearby in scale
The machines closest to Frontier by Rmax, smaller to larger.
- Aurora · USA, operational · 1,012 PFlop/s
- JUPITER · Germany, operational · 1,000 PFlop/s
- HPC7 · Italy, operational · 571.5 PFlop/s
- El Capitan · USA, operational · 1,809 PFlop/s
- LineShine · China, operational · 2,198.4 PFlop/s
Related systems
Same site
- Frontier TDS · USA, operational, 2021
- Gaea · USA, operational, 2023
- Andes (OLCF) · USA, operational, 2020
- Discovery (ORNL) · USA, planned
- Lux (ORNL) · USA, planned
- Summit · USA, decommissioned, 2018
Same operator
- Titan · USA, decommissioned, 2012
- Jaguar · USA, upgraded, 2009
- Phoenix (ORNL) · USA, decommissioned, 2003
- Cheetah (ORNL) · USA, decommissioned, 2002
- Intel Paragon XP/S 150 · USA, decommissioned, 1995
- ORNL Paragon XP/S 35 · USA, operational, 1992
Same accelerator
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 |
|---|---|---|
| hpe.com | does not name this system | Direct liquid cooling, warm water, HPE Cray Operating System, HPE Cray ClusterStor E1000 |
| olcf.ornl.gov | names system and part | AMD Instinct MI250X, Direct liquid cooling, warm water, AMD EPYC 7A53 (Trento), HPE Cray Operating System, HPE Cray ClusterStor E1000 |
Checked 2026-10-18 by pnpm verify. Re-run it and the table changes with the web.
Related news
- TOP500's 66th list confirms JUPITER as the first exascale system outside the US · TOP500El Capitan holds number one at a remeasured 1.809 exaflop/s and Frontier and Aurora hold two and three, so all three US exascale systems in this dataset keep their relative order. JUPITER's 1.000 exaflop/s Booster partition becomes the fourth exascale system worldwide and the first in Europe, confirming the milestone this dataset's earlier rows on Bull's JUPITER delivery had been building toward.
- ORNL, AMD and HPE announce Discovery and Lux as Frontier's AI-era successors · Oak Ridge Leadership Computing FacilityLux runs AMD Instinct MI355X on HPE nodes with Oracle Cloud Infrastructure integration and arrives first, in 2026; Discovery follows in 2028 on MI430X and is billed as beating Frontier on every part of the system, not just peak flops. This keeps ORNL's AMD based lineage going a full hardware generation past the MI300A in El Capitan and the MI250X in Frontier.
- Frontier becomes the first system measured past one exaflop · Oak Ridge Leadership Computing FacilityEvery part of the compute node came from a single supplier, with HPE providing the cabinet, fabric and storage. Two companies account for essentially the whole machine.
Further reading
Inside a node
Why accelerators exist, what the memory hierarchy costs, and how the CPU and accelerator merged onto one package.
Power and cooling
Why air ran out at 20 kW a rack, how warm-water cooling removes the chillers, and what PUE conceals.
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.