Supercomputers/Systems/Bluecrab
USA · operational · hpc
Bluecrab
Also known as Blue Crab
Operated by MARCC at MARCC Data Center (Baltimore) .
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: 1263 of 1345 systems have a readable citation that names them, and 56 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
- 1.40 PFlop/s
- Rmax ÷ Rpeak
- -
- Cores
- 23,280
- 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 servers (R730 GPU nodes, R920 large-memory nodes) | Dell Technologies | - | Reported | web.archive.org |
| Accelerator | - NVIDIA Tesla K80 (two per GPU node) and P100 GPUs | NVIDIA | - | Reported | web.archive.org |
| CPU | - Intel Xeon E5-2680 v3 Haswell, E5-2690 v4 Broadwell, Gold 6126 Skylake and quad-socket E7-8857 v2 Ivy Bridge nodes | Intel | - | Reported | web.archive.org |
| Interconnect | Mellanox InfiniBand FDR Interconnect Fat tree · 56 Gb/s per 4x port · About 1 microsecond FDR-14 InfiniBand with a 2:1 topology | Mellanox NVDANASDAQ
rolls up to NVIDIA | - | Reported | web.archive.org |
| Storage | Lustre Storage Parallel filesystem, separate metadata and object servers 2 PB Lustre (Intel Enterprise Edition for Lustre) plus 14 PB ZFS on Linux | Intel | - | Reported | web.archive.org |
The read
Bluecrab, the main cluster at the Maryland Advanced Research Computing Center in Baltimore, shared by Johns Hopkins researchers. An archived MARCC system description (June 2018 figures) gives over 23,000 cores (23,280 in its table) and over 1.4 PFlop/s combined theoretical performance across Haswell, Broadwell, Skylake and Ivy Bridge nodes on Dell PowerEdge servers: 676 standard Haswell nodes, 50 large-memory quad-socket Ivy Bridge nodes with 1 TB, 48 K80 GPU nodes, Broadwell nodes including 24 K80 nodes and P100 condo nodes, and 48 Skylake nodes, on FDR-14 InfiniBand with 2 PB of Lustre (IEEL) and 14 PB of ZFS on Linux. No LINPACK result is published. The Virtual Institute for I/O lists Bluecrab as the MARCC supercomputer in 2017 and 2021. MARCC's later Rockfish cluster is a separate record.
Timeline
No dated events recorded.
Site & facility
Change history
- 2026-10-09 Added Bluecrab
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 |
|---|---|---|
| web.archive.org | names system and part | Mellanox InfiniBand FDR, Lustre |
| vi4io.org | names system and part | Lustre |
Checked 2026-10-13 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
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.