“As organizations embrace the chances created by generative AI, in addition they should advance sustainability targets, fight escalating energy necessities, and decrease operational prices. The structure we unveiled in the present day makes use of solely liquid cooling, yielding better power and price effectivity benefits than the choice hybrid cooling options in the marketplace,” stated Antonio Neri, president and CEO of HPE, in an announcement.
Supermicro broadcasts liquid cooling providing
Server maker Supermicro referred to as its newest liquid cooling product “an entire liquid cooling resolution.” The providing consists of CDUs, chilly plates, coolant distribution manifolds (CDMs), cooling towers, and end-to-end administration software program.
The liquid cooling resolution consists of chilly plates that dissipate as much as 1600W, sufficient to deal with even the most well liked Nvidia chips. Supermicro’s liquid-cooled SuperClusters are specifically designed for Nvidia‘s GB200 Grace Blackwell Superchip NVL72 platform, which consists of 1 CPU and two Blackwell GPUs on a single die. The clusters characteristic the brand new superior in-rack or in-row coolant distribution models and customized chilly plates designed for the compute tray housing two GB200 chips in a 1U kind issue.
The rack resolution incorporates as much as 72 Blackwell GPUs and 36 Grace CPUs, interconnected by Nvidia’s NVLink community, which might deal with as much as 130 terabytes per second (TB/s) of complete GPU communication with extraordinarily low latency and enhanced efficiency.
The brand new cooling options have began sampling to pick prospects for full-scale manufacturing late within the fourth quarter.
Learn extra about liquid cooling
- Vertiv and Nvidia outline liquid cooling reference structure: Collectively designed structure is meant for GPU-loaded AI factories, which can generate great quantities of warmth.
- Information heart liquid cooling market heats up: Liquid cooling applied sciences for knowledge facilities are transitioning from area of interest choices deployed in particular market segments to mainstream applicability, in line with analysis agency Dell’Oro Group.
- Professionals and cons of air, liquid and geothermal methods: Whether or not it’s to save cash, scale back carbon emissions, adjust to laws or accommodate high-powered AI workloads, enterprises want to function extra energy-efficient knowledge facilities.
- Schneider Electrical shares liquid cooling tips: As high-density knowledge facilities proceed so as to add AI workloads, there’s rising curiosity in liquid cooling, because of its capacity to switch warmth extra effectively than air cooling.
- Information facilities heat as much as liquid cooling: AI, machine studying, and high-performance computing are creating cooling challenges for knowledge heart homeowners and operators. As rack densities enhance and temperatures rise, extra knowledge facilities are discovering methods so as to add liquid cooling to their amenities.
- Accelsius affords liquid cooling with no knowledge heart retrofit: Accelsius, a relative newcomer within the liquid cooling market, has launched its NeuCool dual-phase, direct-to-chip liquid cooling expertise, which is designed to be deployed with out having to do a large retrofit of your knowledge heart.
- ZutaCore launches liquid cooling for superior Nvidia chips: The HyperCool direct-to-chip system from ZutaCore is designed to chill as much as 120kW of rack energy with out requiring a amenities modification.
- Is immersion cooling prepared for mainstream? Liquid cooling began as a fringe expertise however is changing into extra frequent. Proponents hope the identical holds true for immersion cooling.
- Supermicro has a brand new liquid-cooled server for AI: Supermicro’s new server makes use of a self-contained liquid cooling system, and distributors outdoors of the IT infrastructure market are exhibiting curiosity within the expertise.