USA · operational · mixed

Kestrel

Operated by NREL at NREL Golden Campus .

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: -
Rpeak: 44.00 PFlop/s
Rmax ÷ Rpeak: -
Cores: -
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 Cray EX	Hewlett Packard Enterprise HPENYSE	-	Verified	sdxcentral.com fedtechmagazine.com
Accelerator	- NVIDIA H100 GPUs, four per accelerated node; plus NVIDIA A40 GPUs on 8 visualization nodes	NVIDIA NVDANASDAQ	-	Reported	nlr.gov
CPU	- Dual-socket Intel Sapphire Rapids Xeon standard nodes (2,304 nodes, 256 GB each)	Intel INTCNASDAQ	-	Reported	nlr.gov
CPU	AMD EPYC 9004 (Genoa) CPUs x86-64 · 96C · 2.4 GHz · TSMC N5 · 360 W Dual-socket AMD Genoa GPU-node hosts (156 nodes, 384 GB each)	AMD AMDNASDAQ	-	Reported	nlr.gov
Interconnect	HPE Slingshot-11 Interconnect Dragonfly · 200 Gb/s per port · Around 2 microseconds 200 Gb/s HPE Slingshot dragonfly fabric	Hewlett Packard Enterprise HPENYSE	-	Reported	nlr.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
PUE	1.04	2026-02	Operator stated	NLR (ex-NREL) ESIF HPC data centre annualized PUE 1.036 (warm-water cooling). Facility hosting Kestrel, not Kestrel alone. Both pages are NLR.	nlr.gov nlr.gov

The read

HPE Cray EX system at the National Renewable Energy Laboratory (NREL, now the National Laboratory of the Rockies) in Golden, Colorado, replacing the Eagle system for the DOE Office of Energy Efficiency and Renewable Energy. The lab's configuration page lists 2,478 compute nodes: 2,304 dual-socket Intel Sapphire Rapids nodes with 256 GB, 156 dual-socket AMD Genoa nodes with four NVIDIA H100 GPUs each, 10 high-memory Intel nodes and 8 Intel nodes with two NVIDIA A40 GPUs, on a 200 Gb/s HPE Slingshot dragonfly fabric with Lustre scratch and project file systems. SDxCentral cites 44 petaflops of computing capacity, treated here as an operator-stated peak figure; no TOP500 entry is recorded. Reported confidence since node counts have changed across phases of the build (the earlier plan cited 132 GPU nodes).

Timeline

No dated events recorded.

Site & facility

NREL Golden Campus

Location: Golden, CO, USA

sdxcentral.com nlr.gov

Change history

2026-10-01 Added Kestrel (NREL)

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
nlr.gov	names system only	-
sdxcentral.com	names system only	-

Checked 2026-10-09 by pnpm verify. Re-run it and the table changes with the web.