USA · operational · mixed

Sherlock

Operated by Stanford University at Stanford Research Computing 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: 17.70 PFlop/s
Rmax ÷ Rpeak: -
Cores: 76,196
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
Accelerator	NVIDIA Tesla V100 Accelerators CUDA (Volta GV100) · 5,120C · 1.53 GHz · 12 nm · 300 W NVIDIA Tesla V100 (SXM2 and PCIe variants) GPU generation	NVIDIA NVDANASDAQ	-	Reported	sherlock.stanford.edu
Accelerator	NVIDIA H100 SXM5 Accelerators CUDA (Hopper GH100) · 16,896C · 1.98 GHz · TSMC 4N · 700 W NVIDIA Tesla H100 SXM5 GPU generation (Sherlock 4.0)	NVIDIA NVDANASDAQ	-	Reported	sherlock.stanford.edu
Interconnect	Mellanox InfiniBand EDR Interconnect Fat tree · 100 Gb/s per 4x port · About 0.9 microseconds One of three InfiniBand fabrics in use (EDR, HDR, NDR)	Mellanox NVDANASDAQ rolls up to NVIDIA	-	Reported	sherlock.stanford.edu

The read

Stanford's shared condo cluster, available to all Stanford faculty and their research groups, initiated in January 2014 and continuously expanded since; it is now on its fourth hardware generation (Sherlock 4.0). Stanford's own technical specifications page states over 2,000 compute nodes, 76,196 CPU cores and more than 1,200 GPUs spanning generations from Tesla P100 through H100 SXM5 and L40S, aggregating to a vendor-quoted 17.7 PFLOPS of mixed CPU and GPU compute. That figure is not an HPL-measured Linpack result and Sherlock has never been submitted to TOP500, so ranked stays 0 and no rmax_pflops is recorded. The cluster spans three InfiniBand fabrics (EDR, HDR and NDR); Sherlock 4.0 is described as among the first HPC clusters anywhere to offer 400 Gb/s to every node. Most nodes are individually owned by PI groups under a condo model and folded into the shared pool when idle, the same structure UC Berkeley's Savio (also added in this batch) uses. Both sources fetched here are Stanford's own (Stanford Research Computing Center's technical-specifications page and its systems/clusters overview page), so confidence stays reported; no independent vendor case study or press coverage of Sherlock's exact node and GPU counts was found.

Timeline

2014-01 First operational source

Site & facility

Stanford Research Computing Center

Location: Stanford, CA, USA
Commissioned: 2014

sherlock.stanford.edu srcc.stanford.edu

Change history

2026-09-24 Sherlock 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
srcc.stanford.edu	names system only	-
sherlock.stanford.edu	names system and part	NVIDIA Tesla V100, NVIDIA H100 SXM5, Mellanox InfiniBand EDR

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