Supercomputers/Systems/Titan
USA · decommissioned · hpc
Titan
Also known as Cray XK7
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
- 17.59 PFlop/s
- Rpeak
- 27.11 PFlop/s
- Rmax ÷ Rpeak
- 64.9%
- Cores
- 560,640
- Power
- 8.21 MW
- Per watt
- 2.1 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 Titan, 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 | 17.59 PFlop/s | #2 of 193 · top 1% | #34 of 332 · top 10% |
| Energy efficiency | 2.1 GF/W | #29 of 149 · top 19% | #82 of 162 · top 50% |
| Rmax ÷ Rpeak | 64.9 % | #159 of 192 · bottom 18% | #247 of 331 · bottom 26% |
| Power draw | 8,209 kW | #3 of 151 · top 1% | #5 of 171 · top 2% |
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 | - Cray XK7 | Cray HPENYSE
rolls up to Hewlett Packard Enterprise | - | Reported | olcf.ornl.gov |
| Accelerator | NVIDIA Tesla K20X Accelerators CUDA (Kepler GK110) · 2,688C · 0.732 GHz · 28 nm · 235 W NVIDIA K20x | NVIDIA | 18,688 reported | Reported | olcf.ornl.gov |
| CPU | AMD Opteron 6274 (Interlagos) CPUs x86-64 · 16C · 2.2 GHz · 32 nm · 115 W Opteron 6274 16C 2.2GHz | AMD | 18,688 reported | Reported | olcf.ornl.gov |
| Interconnect | Cray Gemini Interconnect 3D torus · About 6 GB/s per direction per link · Around 1 microsecond Cray Gemini interconnect | Cray HPENYSE
rolls up to Hewlett Packard Enterprise | - | Verified | olcf.ornl.govtop500.org |
| Storage | Lustre Storage Parallel filesystem, separate metadata and object servers Spider, Lustre | Cray HPENYSE
rolls up to Hewlett Packard Enterprise | - | Reported | olcf.ornl.gov |
| Cooling | Liquid-assisted air (in-cabinet coil) Cooling In-cabinet air, water-cooled coil · Roughly 20-45 kW per cabinet Cray XK7 cabinet: liquid-assisted air | Cray HPENYSE
rolls up to Hewlett Packard Enterprise | - | Reported | olcf.ornl.gov |
| Operating system | Cray Linux Environment Operating systems Linux, Cray-customised · Proprietary layer over open-source Linux Cray Linux Environment | Cray HPENYSE
rolls up to Hewlett Packard Enterprise | - | Reported | olcf.ornl.gov |
Engineering, funding and access
What Titan 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 | 18,688 nodesOne 16-core AMD Opteron and one NVIDIA K20X Kepler GPU per node, 299,008 CPU cores, 200 cabinets | 2019 | Operator stated | olcf.ornl.gov |
| Total memory | 710 TB32 GB CPU memory plus 6 GB GPU memory per node | 2019 | Operator stated | olcf.ornl.gov |
| Interconnect and topology | Cray Gemini | 2019 | Operator stated | olcf.ornl.gov |
Results
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Notable result | OLCF states Titan delivered 10 times Jaguar's speed and 5 times its energy efficiency in the same footprint | 2012 | Operator stated | olcf.ornl.gov |
The read
Jaguar rebuilt with one NVIDIA K20X per node. The first US No. 1 whose performance came predominantly from GPUs, and the origin of the accelerator-share question this site exists to answer properly: by machine count Titan was one system, but by installed FLOPS it moved a large fraction of US leadership compute onto a single accelerator vendor in one procurement.
Timeline
- 2011 Announced source
- 2012-11 List debut at #1 (November 2012) source
- 2012-11 First operational source
- 2019-08 Retired after seven years as OLCF's flagship system, decommissioned August 1 source
- 2019-08 Decommissioned 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
Where sources disagree
We record conflicts instead of picking a winner quietly. All open questions.
- Decommission date: 2019-08-01 against 2019-08-02. One day apart between two OLCF pages.
Nearby in scale
The machines closest to Titan by Rmax, smaller to larger.
- OpenCanvas GPUaaS SP1 · Japan, operational · 17.53 PFlop/s
- HiPerGator AI · USA, operational · 17.49 PFlop/s
- AOBA-S · Japan, operational · 17.22 PFlop/s
- Pangea III · France, operational · 17.86 PFlop/s
- AI-Farabium · Kazakhstan, operational · 17.93 PFlop/s
- Guru · South Korea, operational · 18 PFlop/s
Related systems
Same site
- Jaguar · USA, upgraded, 2009
- Summit · USA, decommissioned, 2018
- Andes (OLCF) · USA, operational, 2020
- Frontier TDS · USA, operational, 2021
- Phoenix (ORNL) · USA, decommissioned, 2003
- Cheetah (ORNL) · USA, decommissioned, 2002
Same operator
- Frontier · USA, upgraded, 2022
- Discovery (ORNL) · USA, planned
- Lux (ORNL) · USA, planned
- Intel Paragon XP/S 150 · USA, decommissioned, 1995
- ORNL Paragon XP/S 35 · USA, operational, 1992
- Crusher · USA, decommissioned
Same accelerator
- Blue Waters · USA, decommissioned, 2013
- Fionn · Ireland, decommissioned, 2013
- Hydra (MPCDF) · Germany, operational, 2013
- Lichtenberg I · Germany, decommissioned, 2013
- Todi · Switzerland, decommissioned, 2011
- TSUBAME-KFC · Japan, upgraded, 2013
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 |
|---|---|---|
| olcf.ornl.gov | names system, part and count | AMD Opteron 6274 (Interlagos), Lustre · counts: 18688, 18688 |
| olcf.ornl.gov | names system, part and count | AMD Opteron 6274 (Interlagos), Cray Gemini · counts: 18688, 18688 |
Checked 2026-10-18 by pnpm verify. Re-run it and the table changes with the web.
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.