USA · operational · mixed

Savio

Operated by University of California, Berkeley at Berkeley Research Computing Data Center .

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: -
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
Interconnect	Mellanox InfiniBand FDR Interconnect Fat tree · 56 Gb/s per 4x port · About 1 microsecond 56 Gb/s Mellanox FDR InfiniBand, two-level fat tree	Mellanox NVDANASDAQ rolls up to NVIDIA	-	Reported	docs-research-it.berkeley.edu

The read

UC Berkeley's institutional condo cluster, launched in May 2014 and named for Mario Savio of the Free Speech Movement. Berkeley's own Research IT documentation describes a condo model: roughly 40% of nodes are centrally funded for general campus access, with the remaining 60% purchased by individual PI groups ('condo owners') who get priority access to their own nodes while idle cycles flow back to the shared pool, the same structure Stanford's Sherlock (also added in this batch) uses. The current generation, Savio4, runs dual 28-core Intel Xeon Gold 6330 CPUs and offers nodes with eight NVIDIA A5000 or eight NVIDIA L40S GPUs; earlier Savio2 (Haswell) and Savio3 (Skylake, plus Tesla V100/A100/RTX-generation GPUs) hardware remains in service alongside it. The fabric is a 56 Gb/s Mellanox FDR InfiniBand two-level fat tree, with a 12 PB Lustre global scratch filesystem. Berkeley's own overview page also states an April-2020 snapshot of roughly 600 nodes and 15,300 cores at 540 TFLOPS CPU / 1 PFLOPS GPU peak; because that figure is explicitly dated and almost certainly stale given the newer Savio4 hardware described on the same page, no cores or FLOPS figure is recorded on this row. Lawrence Berkeley National Laboratory's ScienceIT page, a legally separate DOE-lab organization from the UC Berkeley campus IT unit, independently describes Berkeley Research Computing's mission for Savio but does not restate its hardware counts, so confidence stays reported rather than verified.

Timeline

2014-05 First operational source

Site & facility

Berkeley Research Computing Data Center

Location: Berkeley, CA, USA
Commissioned: 2014

docs-research-it.berkeley.edu it.lbl.gov

Change history

2026-09-24 Savio added

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
docs-research-it.berkeley.edu	names system and part	Mellanox InfiniBand FDR
it.lbl.gov	names system only	-

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