COMPUTE ATLAS Supercomputer supply-chain graph
1581 systems 558 sites Sourced data

Supercomputers/Systems/Sierra

USA · decommissioned · classified

Sierra

RSS for this system Your watchlist

Also known as ATS-2

Operated by Lawrence Livermore National Laboratory at Livermore Computing Center .

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
94.64 PFlop/s
HPL measured
Rpeak
125.7 PFlop/s
Rmax ÷ Rpeak
75.3%
Cores
1,572,480
Power
7.44 MW
Per watt
12.7 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 Sierra, and those in USA. Computed from our tables; a metric is ranked only when at least eight peers report it.

MetricValueAgainst its eraAgainst its country
Rmax94.64 PFlop/s#3 of 223 · top 1%#12 of 332 · top 3%
Energy efficiency12.7 GF/W#22 of 133 · top 16%#33 of 162 · top 20%
Rmax ÷ Rpeak75.3 %#65 of 216 · top 30%#159 of 331 · top 48%
Power draw7,438 kW#5 of 137 · top 3%#8 of 171 · top 4%

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 Power System AC922
 IBM
IBMNYSE
- Reported hpc.llnl.govolcf.ornl.gov
Accelerator NVIDIA Tesla V100
Accelerators
CUDA (Volta GV100) · 5,120C · 1.53 GHz · 12 nm · 300 W
NVIDIA Volta GV100, 4 per node
 NVIDIA
NVDANASDAQ
17,280
reported
 Reported hpc.llnl.govolcf.ornl.gov
CPU IBM POWER9 22C
CPUs
POWER · 22C · 3.07 GHz · 14 nm · 190 W
IBM POWER9 22C 3.1GHz
 IBM
IBMNYSE
8,640
reported
 Reported hpc.llnl.govolcf.ornl.gov
Interconnect Mellanox InfiniBand EDR
Interconnect
Fat tree · 100 Gb/s per 4x port · About 0.9 microseconds
Dual-rail Mellanox EDR InfiniBand
 Mellanox
NVDANASDAQ rolls up to NVIDIA
- Reported hpc.llnl.govolcf.ornl.gov
Storage IBM Spectrum Scale ESS (GPFS)
Storage
Parallel filesystem (GPFS), erasure-coded declustered RAID · 250 PB on Summit (Alpine) · About 2.5 TB/s
IBM Spectrum Scale (GPFS) at 120 PB
 IBM
IBMNYSE
- Reported hpc.llnl.govasc.llnl.gov
Cooling Direct-to-chip cold plate
Cooling
Cold plate on the package, air for the remainder · Removes roughly 50-80 percent of node heat to liquid
IBM Power System AC922: water-cooled cold plates on CPU and GPU
 IBM
IBMNYSE
- Reported hpc.llnl.govolcf.ornl.gov
Operating system Red Hat Enterprise Linux
Operating systems
Linux distribution · Open source, commercial support subscription
Red Hat Enterprise Linux for POWER
 Red Hat
IBMNYSE rolls up to IBM
- Reported hpc.llnl.govolcf.ornl.gov

Engineering, funding and access

What Sierra runs on, who paid for it and how people get time on it. Each line carries who said it and when.

Hardware

ItemDetailAs ofBasisSource
Compute nodes4,320 nodes4,284 batch and 36 debug nodes; each has 2 IBM POWER9 CPUs and 4 NVIDIA V100 GPUs, 17,280 GPUs in total2024Operator statedhpc.llnl.gov
Interconnect and topologyMellanox EDR InfiniBand2024Operator statedhpc.llnl.gov

Software

ItemDetailAs ofBasisSource
Parallel filesystemRecommended parallel file space /p/gpfs1LLNL page gives the path only2024Operator statedhpc.llnl.gov
Job schedulerLSFQueues pdebug (18 nodes per job, 2 hours) and pbatch (2,048 nodes per job, 24 hours)2024Operator statedhpc.llnl.gov
Programming and software stackRHEL operating system2024Operator statedhpc.llnl.gov

Funding

ItemDetailAs ofBasisSource
Funding programmeAdvanced Simulation and Computing (ASC) program, platform class ATS-2, CORAL-12024Operator statedhpc.llnl.gov
ProcurementCORAL award announced 14 Nov 2014: $325 million to build Summit at ORNL and Sierra at LLNL, both on IBM Power, NVIDIA Volta GPUs and Mellanox interconnectThe $325 million is for both machines combined; expected to be at least seven times more powerful than Sequoia2014-11-14Operator statedscience.osti.gov

The read

Summit's sibling under the same CORAL procurement, with four V100s per node instead of six. The pairing is useful analytically: two systems, same vendors, same year, same fabric, different node ratios: a rare controlled comparison for anyone modelling accelerator-per-node trends. Stockpile stewardship workload.

Timeline

  • 2014 Announced source
  • 2014-11-14 DOE announces the CORAL award of $325 million to build Sierra at LLNL and Summit at ORNL, with Sierra expected to be at least seven times more powerful than Sequoia source
  • 2018-06 List debut at #3 (June 2018) source
  • 2018-06 First operational source
  • 2025-09-30 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

Livermore Computing Center

Location
Livermore, CA, USA
Cooling
Direct liquid cooling (warm water)
Commissioned
1996

Change history

Nearby in scale

The machines closest to Sierra by Rmax, smaller to larger.

  • Sunway TaihuLight · China, operational · 93.01 PFlop/s
  • CHIE-3 · Japan, operational · 91.94 PFlop/s
  • CEA-HE · France, operational · 90.79 PFlop/s
  • Venado · USA, operational · 98.51 PFlop/s
  • SSC-24 · South Korea, operational · 106.2 PFlop/s
  • MAXIMUS-384 · USA, operational · 114.5 PFlop/s

Same site

Same operator

  • Ruby (LLNL) · USA, operational, 2020
  • rzVernal · USA, operational, 2022
  • Tenaya · USA, operational, 2022
  • Tioga · USA, operational, 2022
  • Bengal · USA, operational, 2023
  • Dane · USA, operational, 2023

Same accelerator

  • ABCI · Japan, decommissioned, 2018
  • Circe · USA, operational, 2018
  • CLAIX-2018 · Germany, decommissioned, 2018
  • COBRA · Germany, decommissioned, 2018
  • Govorun · Russia, upgraded, 2018
  • JUWELS Cluster · Germany, operational, 2018

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 sourceResultFound on the page
hpc.llnl.govnames system only-
olcf.ornl.govdoes not name this systemIBM POWER9 22C, Mellanox InfiniBand EDR, IBM Spectrum Scale ESS (GPFS)

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.