Supercomputers/Systems/VISION (TAMUS)
USA · operational · mixed
VISION (TAMUS)
Also known as TAMUS VISION supercomputer
Operated by Texas A&M University System .
Researched with AI assistance, checked by scripts. These rows were researched by AI agents reading public operator, laboratory and vendor sources, appended by scripts and checked by automated validation; no subject-matter expert has reviewed them (how this site is made). A mechanical second-reader pass has since fetched every cited source: 1350 of 1581 systems have a readable citation that names them, and 64 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
- 34.82 PFlop/s
- Rpeak
- 51.16 PFlop/s
- Rmax ÷ Rpeak
- 68.1%
- Cores
- 110,960
- 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.
How it compares
Ranked against the other machines in this dataset that report the same figure: those that first came into service within two years of VISION (TAMUS), and those in USA. Computed from our tables; a metric is ranked only when at least eight peers report it.
| Metric | Value | Against its era | Against its country |
|---|---|---|---|
| Rmax | 34.82 PFlop/s | #58 of 202 · top 28% | #19 of 332 · top 5% |
| Rmax ÷ Rpeak | 68.1 % | #127 of 199 · bottom 37% | #215 of 331 · bottom 36% |
Every metric is ranked largest-first, so a high power-draw rank means a big power draw, not a good one. Rmax ÷ Rpeak uses HPL-measured results only. Rankings move as machines are added; they describe this dataset, not the world.
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 | - 95 NVIDIA DGX H200 nodes in a 3 scalable-unit SuperPOD | NVIDIA | 95 reported | Reported | docs.vision.tamus.edutop500.org |
| Accelerator | NVIDIA H200 SXM Accelerators CUDA (Hopper GH100) · 16,896C · 1.98 GHz · TSMC 4N · 700 W Eight NVIDIA H200 GPUs per DGX node, 760 GPUs total (720 schedulable) | NVIDIA | 760 reported | Reported | docs.vision.tamus.edutop500.org |
| CPU | Intel Xeon Platinum 8480C CPUs x86-64 · 56C · 2 GHz · Intel 7 · 350 W Two Intel Xeon Platinum 8480C (56 cores) per DGX node across 95 nodes | Intel | 190 estimated | Reported | docs.vision.tamus.edutop500.org |
| Interconnect | NVIDIA InfiniBand NDR (Quantum-2) Interconnect Fat tree or dragonfly+ · 400 Gb/s per port · Under 0.6 microseconds 400G InfiniBand compute and storage fabrics with 54 Quantum switches | NVIDIA | - | Reported | docs.vision.tamus.edutop500.org |
| Storage | DDN EXAScaler Storage Lustre appliance DDN EXAScaler, 8 PB fast scratch | DDN | - | Reported | docs.vision.tamus.edu |
| Storage | - 1 PB Pure Storage network file system | Pure Storage | - | Reported | docs.vision.tamus.edu |
| Operating system | Ubuntu Server Operating systems Linux distribution · Open source, free with paid support tiers Ubuntu 24.04 | Canonical | - | Reported | top500.org |
Engineering, funding and access
What VISION (TAMUS) runs on, who paid for it and how people get time on it. Each line carries who said it and when.
Hardware
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Compute nodes | 95 nodes95 NVIDIA DGX nodes with 8 H200 GPUs each (760 GPUs); 90 nodes and 720 GPUs are schedulable; 2 login nodes | 2025 | Operator stated | docs.vision.tamus.edu |
| Total memory | 190 TB2 TB DDR5 per DGX node across 95 nodes; each H200 GPU also has 144 GB | 2025 | Operator stated | docs.vision.tamus.edu |
| Storage capacity | 8 PBDDN EXAScaler fast scratch with 4.8 TB/s total throughput | 2025 | Operator stated | docs.vision.tamus.edu |
| Interconnect and topology | NVIDIA Quantum InfiniBand at 400 Gb/s on separate compute and storage fabrics (54 switches), plus Ethernet spine-leaf with 12 Cumulus switches | 2025 | Operator stated | docs.vision.tamus.edu |
Software
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Parallel filesystem | DDN EXAScaler, 19 storage nodes, 8 PB fast scratch, 4.8 TB/s | 2025 | Operator stated | docs.vision.tamus.edu |
| Programming and software stack | NVIDIA H200 GPUs with MIG partitioning into up to 7 instances per GPU (5,040 MIG instances schedulable) | 2025 | Operator stated | docs.vision.tamus.edu |
The read
The Texas A&M University System's NVIDIA DGX SuperPOD (a 3 scalable-unit reference build), open to researchers at every TAMUS institution and in general availability. The TAMUS documentation lists 95 DGX H200 nodes (90 schedulable), each with eight H200 GPUs and two Xeon Platinum 8480C CPUs (760 GPUs in total, 720 schedulable), 400G InfiniBand compute and storage fabrics with 54 Quantum switches, an 8 PB DDN EXAScaler scratch system and a 1 PB Pure Storage NFS tier; the project site says it triples the System's previous supercomputing capacity. TOP500 (June 2026, 66th; list full for our table so no list row) lists NVIDIA DGX H200 with Xeon Platinum 8480C, HGX H200 141 GB, InfiniBand NDR and Ubuntu 24.04, installed 2025, with 110,960 cores, Rmax 34.82 and Rpeak 51.16 PFlop/s. The TAMUS spec page labels 34.82 PFlop/s as 'Peak DP FLOPS', which matches TOP500's measured Rmax rather than its Rpeak, and TAMUS's 10,640 CPU cores differ from TOP500's 110,960 core count (TOP500 appears to count GPU cores too, our inference). The fetched pages do not say where the machine is housed, so site is blank.
Timeline
- 2025 First operational source
Change history
- 2026-10-17 Added VISION (TAMUS)
Nearby in scale
The machines closest to VISION (TAMUS) by Rmax, smaller to larger.
- Sunrise · United Kingdom, operational · 34.77 PFlop/s
- SAKURAONE · Japan, operational · 33.95 PFlop/s
- Sejong · South Korea, operational · 32.97 PFlop/s
- HPC5 · Italy, operational · 35.45 PFlop/s
- Shaheen III · Saudi Arabia, operational · 35.66 PFlop/s
- SAKURAONE CYD · Japan, operational · 37.79 PFlop/s
Related systems
Same accelerator
- ABCI 3.0 · Japan, operational, 2025
- AI-Farabium · Kazakhstan, operational, 2025
- Alem.Cloud · Kazakhstan, operational, 2025
- AzInTelecom Supercomputer Center · Azerbaijan, operational
- Cannon2-H200 · USA, operational, 2025
- davinci-2 · Italy, operational, 2025
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.vision.tamus.edu | names system, part and count | NVIDIA H200 SXM, Intel Xeon Platinum 8480C, NVIDIA InfiniBand NDR (Quantum-2) · counts: 760 |
| top500.org | names system and part | NVIDIA H200 SXM, Intel Xeon Platinum 8480C, NVIDIA InfiniBand NDR (Quantum-2), Ubuntu Server |
Checked 2026-10-18 by pnpm verify. Re-run it and the table changes with the web.
Further reading
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.
Inside a node
Why accelerators exist, what the memory hierarchy costs, and how the CPU and accelerator merged onto one package.
The interconnect
Topologies, why latency and tail behaviour matter more than bandwidth, and how the fabric market consolidated.