IBM Blue Vela

Operated by IBM .

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
Integrator	- Dell PowerEdge XE9680 nodes on the NVIDIA H100 SuperPOD reference architecture	Dell Technologies DELLNYSE	-	Reported	blocksandfiles.com arxiv.org
Accelerator	NVIDIA H100 SXM5 Accelerators CUDA (Hopper GH100) · 16,896C · 1.98 GHz · TSMC 4N · 700 W 5,000 NVIDIA H100 80GB GPUs, eight per node	NVIDIA NVDANASDAQ	5,000 reported	Reported	blocksandfiles.com arxiv.org
CPU	- Dual 48-core 4th Gen Intel Xeon Scalable per node, 2TB RAM	Intel INTCNASDAQ	-	Reported	arxiv.org
Interconnect	NVIDIA InfiniBand NDR (Quantum-2) Interconnect Fat tree or dragonfly+ · 400 Gb/s per port · Under 0.6 microseconds Separate compute and storage InfiniBand fabrics, non-blocking rail-optimised fat tree, ConnectX-7 adapters	NVIDIA NVDANASDAQ	-	Reported	arxiv.org blocksandfiles.com
Storage	- IBM Storage Scale System 6000, two chassis, almost 3 PB raw NVMe	IBM IBMNYSE	-	Reported	blocksandfiles.com arxiv.org

The read

IBM Research's on-premises H100 training cluster, built with Dell and NVIDIA on the H100 SuperPOD reference architecture: Dell PowerEdge XE9680 nodes with eight 80GB H100 GPUs and dual 48-core fourth-generation Xeon CPUs, in 32-node scalable units, four to a 128-node compute pod, on non-blocking rail-optimised InfiniBand with separate compute and storage fabrics. Storage is IBM Storage Scale System 6000, almost 3 PB of raw NVMe. IBM's own paper (January 2025 revision) says Blue Vela began coming online in April 2024, that it lifted IBM Research's GPU capacity 104 percent over 2023 and that it was to reach a cumulative 214 percent by the end of 2024. The 5,000-GPU figure comes from Blocks and Files summarising IBM material, not from a page IBM publishes with the count, so this stays reported. IBM says only that the hosting site runs on 100 percent renewable energy; no location or power figure is given, so USA records only the operator's home country. It has never submitted an HPL result.

Timeline

2024-04 First operational source

Change history

2026-10-02 IBM Blue Vela added (5,000 H100 on-premises training cluster)

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
arxiv.org	does not name this system	NVIDIA H100 SXM5, NVIDIA InfiniBand NDR (Quantum-2)
blocksandfiles.com	names system, part and count	NVIDIA H100 SXM5, NVIDIA InfiniBand NDR (Quantum-2) · counts: 5000

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

IBM Blue Vela

Measured performance

What this machine is made of

The read

Timeline

Change history

Source check

Further reading

Inside a node

The interconnect

Storage and I/O