Supercomputers/Systems/Insomnia (Columbia)
USA · operational · mixed
Insomnia (Columbia)
Also known as Columbia Insomnia HPC cluster
Operated by Columbia University .
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
- 7,144
- 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 A6000 GPUs (page writes A600): 13 nodes with 8 and 2 nodes with 4 | NVIDIA | 112 reported | Reported | columbiauniversity.atlassian.net |
| Accelerator | - 3 GPU nodes with two NVIDIA H100 GPUs each | NVIDIA | 6 reported | Reported | columbiauniversity.atlassian.net |
| Accelerator | - 3 GPU nodes with two NVIDIA L40 GPUs each | NVIDIA | 6 reported | Reported | columbiauniversity.atlassian.net |
| Accelerator | NVIDIA L40S Accelerators CUDA (Ada Lovelace) · 350 W 9 GPU nodes with two NVIDIA L40S GPUs each | NVIDIA | 18 reported | Reported | columbiauniversity.atlassian.net |
| CPU | Intel Xeon Platinum 8460Y+ CPUs x86-64 · 40C · 2 GHz · Intel 7 · 300 W Dual Intel Xeon Platinum 8460Y processors (2 GHz) in every server | Intel | - | Reported | columbiauniversity.atlassian.net |
| Interconnect | NVIDIA InfiniBand HDR Interconnect Fat tree or dragonfly+ · 200 Gb/s per port · About 0.6 microseconds HDR InfiniBand | NVIDIA | - | Reported | columbiauniversity.atlassian.net |
| Operating system | Red Hat Enterprise Linux Operating systems Linux distribution · Open source, commercial support subscription Red Hat Enterprise Linux 9.3 | Red Hat | - | Reported | columbiauniversity.atlassian.net |
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 |
|---|---|---|---|---|---|
| Accelerators | 142 accelerators | 2026-10 | Operator stated | Sum of the listed GPU node types (112 A6000-class, 6 H100, 6 L40, 18 L40S). | columbiauniversity.atlassian.net |
The read
Columbia's shared HPC cluster, which Columbia says went live in February 2024 as a joint purchase by 21 research groups and departments. The page lists 90 nodes with dual Intel Xeon Platinum 8460Y CPUs (7,144 physical cores): 41 standard, 19 high-memory and 30 GPU nodes. GPU nodes are 13 with eight A6000-class GPUs (the page writes the model as A600), 2 with four, 3 with two H100, 3 with two L40 and 9 with two L40S, which sums to 142 GPUs by our arithmetic. Interconnect is HDR InfiniBand with 2.282 PB of GPFS. No FLOPS figure is published.
Timeline
- 2024-02 First operational source
Change history
- 2026-10-07 Added Insomnia (Columbia)
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 |
|---|---|---|
| columbiauniversity.atlassian.net | names system only | · counts: 18 |
Checked 2026-10-09 by pnpm verify. Re-run it and the table changes with the web.
Further reading
Inside a node
Latency-optimised versus throughput-optimised processors, NUMA, and the end of the host-to-device copy.
The interconnect
Topologies, why latency and tail behaviour matter more than bandwidth, and how the fabric market consolidated.
Programming a supercomputer
The software stack beneath the application, and why the OS is now a sovereignty signal as much as a technical choice.