Skip to content

Loring Data Center / An HPC data center for research and universities

An HPC data center for research and universities

Expand the cluster without building the room. Dense CPU compute on liquid-cooled racks, with the heat carried away in liquid instead of fought with air.

Request capacityHow the cooling works
Exterior of Building 7230 at Loring Commerce Centre
Building 7230, Loring Commerce Centre.
24racks doing the work of 115 air-cooled racks
4 kWper rack unit in a 2U LiquidCool ZPServer
1.13PUE, against about 2.25 air-cooled
5 to 15 msfiber latency to Boston and Halifax

Who it is for

  • Universities and research computing groupsRoom for the next cluster when the campus machine room is out of power or cooling.
  • Hospital and medical researchDense compute for imaging, genomics and modeling, with a second copy of data kept off campus.
  • Engineering and simulation teamsManufacturers and engineering firms running simulation and modeling workloads that outgrew the office.
  • Labs and research institutesSteady, high-utilization compute that runs all year, where operating cost matters more than metro latency.

Why liquid suits HPC

  • Coolest fluid to the hottest partsDirected flow sends fresh coolant to the processors first, then the liquid bathes the rest of the electronics.
  • Density without the noiseNo fans in the servers and no air handlers in the hall. The compute of 115 air-cooled racks fits in 24.
  • Standard hardwareOff-the-shelf Intel and AMD platforms, standard memory and drives, standard 19 inch racks, serviced the way you do now with dripless quick disconnects.
  • Lower running costClose to 90 percent of each kilowatt reaches the servers, on hydro-backed power priced well below Boston-area rates.

What to bring

  • Your current specNode count, processors, memory, storage and interconnect. Loring packages comparable platforms in LiquidCool chassis, or you bring immersion-ready hardware.
  • GPU workLoring is focused on dense CPU compute. GPU deployments are considered case by case.

Questions

Can Loring host an HPC cluster?

Yes. Loring is built around dense, liquid-cooled CPU compute. Send the node spec and Loring will tell you what fits, packaged in LiquidCool chassis or on your own immersion-ready hardware.

Do you host GPU clusters?

Loring is focused on dense CPU compute, storage and disaster recovery. GPU deployments are considered case by case.

Is the latency good enough for research users?

Fiber latency is 5 to 15 ms to Boston and Halifax, which suits batch and steady-state research computing rather than interactive trading-style workloads.

Tell us what you run.

Email info@loringdc.com or call (207) 300-5046. The email opens with six questions filled in. The power figure and the date matter most.

Or use the capacity form. Site visits by appointment at Building 7230, 89 Kansas Road, Limestone, Maine.