Supercomputers/Systems/SEAS H100 (Harvard)
USA · operational · hpc
SEAS H100 (Harvard)
Also known as Harvard SEAS H100
Operated by Harvard FAS Research Computing .
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
- 3.38 PFlop/s
- Rpeak
- 4.38 PFlop/s
- Rmax ÷ Rpeak
- 77.2%
- Cores
- 13,104
- Power
- 80.55 kW
- Per watt
- 42.0 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 SEAS H100 (Harvard), 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 | 3.38 PFlop/s | #197 of 234 · bottom 17% | #105 of 332 · top 31% |
| Energy efficiency | 42.0 GF/W | #57 of 121 · top 47% | #19 of 162 · top 11% |
| Rmax ÷ Rpeak | 77.2 % | #100 of 231 · top 43% | #139 of 331 · top 42% |
| Power draw | 81 kW | #117 of 126 · bottom 8% | #163 of 171 · bottom 6% |
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 | - Lenovo ThinkSystem SD665-N V3 | Lenovo | - | Reported | top500.org |
| Accelerator | NVIDIA H100 SXM5 Accelerators CUDA (Hopper GH100) · 16,896C · 1.98 GHz · TSMC 4N · 700 W NVIDIA H100 80GB | NVIDIA | - | Reported | top500.org |
| CPU | AMD EPYC 9004 (Genoa) CPUs x86-64 · 96C · 2.4 GHz · TSMC N5 · 360 W AMD EPYC 9454 48C 2.75GHz | AMD | - | Reported | top500.org |
| Interconnect | NVIDIA InfiniBand HDR Interconnect Fat tree or dragonfly+ · 200 Gb/s per port · About 0.6 microseconds Infiniband HDR | NVIDIA | - | Reported | top500.org |
The read
H100 GPU nodes of Harvard's School of Engineering and Applied Sciences, listed by TOP500 as SEAS H100: Lenovo ThinkSystem SD665-N V3 with AMD EPYC 9454 48-core 2.75 GHz, NVIDIA H100 80GB and InfiniBand HDR under Linux, installed 2024. It has 13,104 cores, Rmax 3.38 and Rpeak 4.38 PFlop/s at 80.55 kW (power measurement level 1), and debuted at 298th in November 2024 and was 392nd in June 2026. FAS Research Computing's SEAS page describes the seas_gpu partition as 244 GPUs 'ranging from Nvidia A100 to Nvidia H200' with 3,904 cores 'ranging from Intel Ice Lake to AMD Genoa', so it supports the context (Genoa-class AMD hosts and NVIDIA GPUs inside the shared Cannon environment) but names neither H100, Lenovo nor a node count, and the TOP500 figures are the only machine-specific numbers. Harvard's SEAS wiki says the remaining SEAS hardware is in maintenance mode and points users to FAS Research Computing, which is why the operator is recorded as FASRC. Other Harvard systems are harvard-cannon and harvard-cannon2-h200; the tier is reported.
Timeline
Public list appearances
Pointers only: rank and edition, linking to the canonical entry. We do not reproduce list tables. See sourcing policy.
Change history
- 2026-10-18 Added SEAS H100 (Harvard)
Nearby in scale
The machines closest to SEAS H100 (Harvard) by Rmax, smaller to larger.
- ASPIRE 2A (GPU Partition) · Singapore, operational · 3.33 PFlop/s
- SEIRAN · Japan, operational · 3.32 PFlop/s
- Topaz (ERDC) · USA, operational · 3.32 PFlop/s
- Tenaya · USA, operational · 3.41 PFlop/s
- davinci-1 · Italy, operational · 3.45 PFlop/s
- CARO · Germany, operational · 3.46 PFlop/s
Related systems
Same operator
- Cannon2-H200 · USA, operational, 2025
- Cannon (FASRC) · USA, upgraded, 2019
- Harvard Odyssey · USA, operational, 2008
Same accelerator
- ARF-ACC · Turkey, operational, 2024
- Arka AI/ML · India, operational, 2024
- ASPIRE 2A+ · Singapore, operational, 2024
- BioHive-2 · USA, operational, 2024
- CHIE-2 · Japan, operational, 2024
- CHIE-3 · Japan, operational, 2024
Further reading
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.
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.