Supercomputers/Systems/Roadrunner
USA · decommissioned · classified
Roadrunner
Operated by Los Alamos National Laboratory at Los Alamos Strategic Computing Complex .
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
- 1.03 PFlop/s
- Rpeak
- 1.38 PFlop/s
- Rmax ÷ Rpeak
- 74.6%
- Cores
- 122,400
- Power
- 2.35 MW
- Per watt
- 0.4 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.
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 Roadrunner, 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 | 1.03 PFlop/s | #8 of 125 · top 6% | #180 of 332 · bottom 47% |
| Energy efficiency | 0.4 GF/W | #9 of 71 · top 11% | #133 of 162 · bottom 19% |
| Rmax ÷ Rpeak | 74.6 % | #94 of 125 · bottom 26% | #167 of 331 · top 50% |
| Power draw | 2,345 kW | #10 of 72 · top 13% | #25 of 171 · top 14% |
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 | - IBM BladeCenter QS22/LS21 | IBM | - | Reported | ibm.comlanl.gov |
| Accelerator | IBM PowerXCell 8i Accelerators Cell BE (PPE + 8 SPE) · 9C · 3.2 GHz · 65 nm · 92 W IBM PowerXCell 8i | IBM | - | Reported | ibm.comlanl.gov |
| CPU | AMD Opteron 2210 (Santa Rosa) CPUs x86-64 · 2C · 1.8 GHz · 90 nm · 95 W AMD Opteron dual-core, hybrid tri-blade | AMD | - | Estimated | ibm.comlanl.gov |
| Interconnect | Mellanox InfiniBand DDR Interconnect Fat tree · 20 Gb/s per 4x port · About 2 microseconds InfiniBand DDR | Mellanox NVDANASDAQ
rolls up to NVIDIA | - | Reported | ibm.comlanl.gov |
| Storage | Panasas ActiveScale (PanFS) Storage Object storage, blade shelves (1 director + 10 storage blades) Comprised of Panasas blades RAID 5 or RAID 10 on a per-object basis | Panasas | - | Verified | msstconference.orgroute-fifty.com |
| Cooling | Air cooling (CRAC/CRAH) Cooling Room-level air handling · Practical ceiling around 15-20 kW per rack Roadrunner is air-cooled utilizing large copper heat sinks and variable-speed fans | IBM | - | Reported | linuxjournal.com |
| Operating system | Red Hat Enterprise Linux Operating systems Linux distribution · Open source, commercial support subscription Roadrunner runs on Linux - Red Hat to be exact | Red Hat | - | Reported | linuxjournal.com |
Engineering, funding and access
What Roadrunner runs on, who paid for it and how people get time on it. Each line carries who said it and when.
Software
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Programming and software stack | Linux on a hybrid design pairing Opteron dual-core processors with PowerXCell 8i accelerators | 2010 | Operator stated | osti.gov |
Funding
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Funded by | DOE National Nuclear Security Administration (NNSA) Advanced Simulation and Computing program | 2013 | Operator stated | web.ornl.gov |
| Stated cost | 120 USD mAbout USD 120 million for all project phases. | 2010 | Operator stated | osti.gov |
| Procurement | IBM built it with Los Alamos and NNSA; Cell collaboration began 2002, work on Roadrunner 2006, Phase 1 was a 71 TFlop/s Opteron cluster, Phase 2 completed October 2007 | 2010 | Operator stated | osti.gov |
Results
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Notable result | No. 1 on the TOP500 from June 2008 into 2009 and No. 3 on the Green500; helped close a gap in understanding energy flow in a weapon (2008) | 2008-06 | Operator stated | energy.gov |
| Notable result | First supercomputer to sustain one petaflop/s on Linpack, 1.026 petaflop/s at 74.5 percent efficiency on 26 May 2008 (2008) | 2008-05 | Operator stated | osti.gov |
The read
The first petaflop machine, and the first heterogeneous one: AMD Opteron host processors paired with IBM PowerXCell 8i accelerators in a hybrid tri-blade. The accelerator/host split it introduced is now the default architecture for essentially everything on this site. The Cell line itself did not survive: an early demonstration that being first in the accelerator role does not confer durable share.
Timeline
- 2006 Announced source
- 2007-10 Phase 2 was completed, with the Opteron-only Phase 1 cluster of 71 teraflop/s already in production source
- 2008-05 Sustained 1.026 petaflop/s on Linpack on 26 May, the first computer to exceed one petaflop/s source
- 2008-06 List debut at #1 (June 2008) source
- 2008-06 First operational source
- 2013-03 Decommissioned on March 31 due to low energy efficiency compared to newer systems source
- 2013-03 Decommissioned source
- 2013-04 After decommissioning the machine was wiped and declassified, giving researchers about a month of experiments before dismantling 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
Los Alamos Strategic Computing Complex
- Location
- Los Alamos, NM, USA
- Cooling
- Chilled water
- Commissioned
- 1993
Where sources disagree
We record conflicts instead of picking a winner quietly. All open questions.
- Processor and node count: 6,563 dual-core Opterons each linked to a PowerXCell 8i against 6,948 tri-blade nodes with 13,896 x86 cores. The two counts differ by roughly a factor of two; no node count is shown on the page until one is confirmed.
Nearby in scale
The machines closest to Roadrunner by Rmax, smaller to larger.
- Lonestar 5 · USA, decommissioned · 1.02 PFlop/s
- QUARTETTO · Japan, operational · 1.02 PFlop/s
- MareNostrum P9 CTE · Spain, operational · 1.02 PFlop/s
- Lengau · South Africa, operational · 1.03 PFlop/s
- Taurus · Germany, decommissioned · 1.03 PFlop/s
- TX-Green · USA, operational · 1.03 PFlop/s
Related systems
Same site
- Cielo · USA, decommissioned, 2011
- Mustang (LANL) · USA, decommissioned, 2011
- ASCI Q · USA, decommissioned, 2002
- Grizzly · USA, operational, 2016
- Trinity · USA, decommissioned, 2017
- ASCI Blue Mountain · USA, decommissioned, 1998
Same operator
- Cerrillos · USA, decommissioned, 2008
- LANL Lightning · USA, operational, 2003
- CM-5/1024 · USA, decommissioned, 1993
- Crossroads · USA, operational, 2023
- Venado · USA, operational, 2024
- Mission (LANL) · USA, planned
Same accelerator
- Lomonosov · Russia, decommissioned, 2009
- QPACE (Julich) · Germany, operational, 2009
- QPACE (Wuppertal) · Germany, operational, 2009
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 |
|---|---|---|
| energy.gov | names system only | - |
| osti.gov | could not be fetched | - |
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.
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.
The interconnect
Topologies, why latency and tail behaviour matter more than bandwidth, and how the fabric market consolidated.