Home » SUPERMICRO Announces Rack-Scale Plug-and-Play Liquid-Cooled AI SuperCluster Powered by NVIDIA Blackwell and NVIDIA HGX H100/H200 |

SUPERMICRO Announces Rack-Scale Plug-and-Play Liquid-Cooled AI SuperCluster Powered by NVIDIA Blackwell and NVIDIA HGX H100/H200 |

by admin
SUPERMICRO Announces Rack-Scale Plug-and-Play Liquid-Cooled AI SuperCluster Powered by NVIDIA Blackwell and NVIDIA HGX H100/H200 |

A liquid-cooled answer that brings innovation to the AI ​​period – The SuperCluster for generative AI, built-in with NVIDIA AI Enterprise and NIM microservices, will increase the return on funding, improves AI effectivity, and offers monstrous computing models for Fulfillment Fast transport.

SAN JOSE, Calif. and TAIPEI, June 6, 2024 /PRNewswire/ — Super Micro Computer, Inc.(NASDAQ: SMCI)is a supplier of full IT options resulting in AI, cloud, HPC, storage, 5G/Edge, and inexperienced computing. The firm introduced SuperCluster, a cloud-based answer that improves productive AI growth and deployment of the NVIDIA AI Enterprise software program platform and accelerates the adoption of productive AI throughout industries and companies. Supermicro’s 4U liquid-cooled system, which may be shortly deployed in liquid-cooled AI information facilities, delivers full AI efficiency with NVIDIA’s lately introduced Blackwell GPU, which delivers 20 petaflops per GPU for coaching efficiency and an estimated 30 instances extra efficiency in comparison with earlier era GPUs, contributing to additional price reductions over earlier merchandise. Building on its technique of staying forward of the market with new merchandise and applied sciences, Supermicro lately introduced an entire lineup of merchandise primarily based on the NVIDIA Blackwell structure with assist for the NVIDIA HGXTM B100, B200, GB200 and Grace Blackwell Superchip.

Plug-and-Play Liquid-Cooled AI SuperCluster

Charles Liang, President and CEO of Supermicro, stated: “Supermicro continues to guide the trade within the manufacturing and supply of liquid-cooled AI options. You get essentially the most on your cash with NVIDIA AI Enterprise software program. This shortens the supply time of turnkey cluster methods with choices for liquid cooling (water cooling) or air cooling, together with NVIDIA HGX H100 and H200 and upcoming B100, B200, and GB200 tower options, our custom-made liquid cooling options can cut back your information heart’s ongoing power consumption by as much as 40%.

Please see right here for particulars:www.supermicro.com/ai

See also  Saudi Arabia and Iran resume diplomatic relations. With the help of Beijing

At COMPUTEX 2024, Supermicro will announce upcoming methods optimized for NVIDIA Blackwell GPUs, together with 10U air-cooled and 4U liquid-cooled NVIDIA HGX B200 methods. In addition, Supermicro presents the NVIDIA HGX B100 air-cooled 8U system and Supermicro’s NVIDIA GB200 NVL72 rack with 72 GPUs related by way of NVIDIA NVLink change, NVIDIA H200 NVL PCIe GPU, and NVIDIA GB200 NVL2 NVL2 lately introduced. methods assist structure.

NVIDIA founder and CEO Jensen Huang stated: “Generative AI advocates resetting your complete information stack GPU-accelerated and designed for AI. We design computing and networking options that allow $1 billion world information facilities for the age of AI.”

With the fast growth of main language fashions and the continued introduction of latest open supply fashions resembling Meta’s Llama-3 and Mistral’s Mixtral 8x22B, immediately’s AI fashions have gotten extra inexpensive for companies. To preserve tempo with the present fast evolution of the AI ​​revolution, you will need to simplify and supply the accessibility of AI infrastructure in an economical method. Supermicro’s cloud-native AI SuperCluster makes use of NVIDIA AI Enterprise to bridge the hole between cloud usability and portability, permitting you to seamlessly transfer AI tasks from pilot to manufacturing at any scale. This offers you the flexibility to deploy anyplace, with securely managed information, together with self-hosted methods or massive on-premises information facilities.

As corporations throughout industries quickly take a look at manufacturing AI use instances, Supermicro is working intently with NVIDIA to assist convey AI purposes from testing and piloting to manufacturing, and deploying a big AI information heart permits a seamless and versatile transition Lo the result’s achieved by way of rack- and cluster-level growth utilizing the NVIDIA AI Enterprise software program platform, enabling a clean transition from preliminary testing to fast AI implementation.

Managed providers can implement consensus in selecting infrastructure, sharing information, and managing productive AI methods. However, NVIDIA NIM microservices, that are a part of the NVIDIA AI Enterprise platform, supply the advantages of each AI managed era and open supply deployment. Its versatile runtime accelerates AI mannequin deployment on a variety of fashions, from open supply to NVIDIA core fashions. Additionally, NVIDIA NeMoTM permits customized mannequin growth with information choice, superior customization, and Retrieval-Augmented Generation (RAG) for business-ready options. Combined with Supermicro’s NVIDIA AI Enterprise Ready SuperCluster, NVIDIA NIM offers the quickest strategy to deploy quicker, extra productive AI.

See also  FrieslandCampina fined €561,000 for infant formula

SuperCluster’s present AI-powered productiveness merchandise by Supermicro embrace:

Liquid cooling (water cooling) sort Supermicro NVIDIA HGX H100/H200 SuperCluster: Scale computing unit with 256 H100/H200 GPUs put in in 5 racks (together with one devoted community rack).
the air is coolSupermicro NVIDIA HGX H100/H200 SuperCluster: Scalable computing unit with 256 HGX H100/H200 GPUs put in in 9 racks (together with one devoted community rack).
Supermicro NVIDIA MGX GH200 SuperCluster: Expandable computing unit with 256 GH200 GraceTM Hopper Superchips in 9 racks (together with one devoted community rack)

Supermicro SuperCluster is NVIDIA AI Enterprise enabled and makes use of NVIDIA NIM microservices and the NVIDIA NeMo platform to offer end-to-end AI customization. It can be optimized for NVIDIA Quantum-2 InfiniBand and the brand new NVIDIA Spectrum-X Ethernet platform with community speeds of 400Gb/s per GPU, permitting you to entry massive clusters with tens of hundreds of GPUs.

Upcoming SuperCluster merchandise embrace:

Supermicro NVIDIA HGX B200 SuperCluster: Liquid cooling (water cooling) sort
Supermicro NVIDIA HGX B100/B200 SuperCluster: the air is cool
Supermicro NVIDIA GB200 NVL72 or NVL36 SuperCluster: Liquid cooling (water cooling) sort

Supermicro’s SuperCluster answer is designed for LLM coaching, deep studying, and enormous batch dimension profiling. Supermicro’s L11 and L12 certification testing and on-site deployment providers present a hassle-free expertise. Plug-and-play, scalable models simplify information heart deployments and cut back turnaround time.

Super Micro Computer, Inc. about

Supermicro (NASDAQ: SMCI) is a world chief in custom-made {hardware} and full IT options. Founded and primarily based in San Jose, Calif., Supermicro is dedicated to delivering early innovation to the IT infrastructure market, together with enterprise, cloud, AI, HPC, and IoT/Edge. We are a producer of full IT options, offering servers, AI, storage, IoT, community switches, software program, and upkeep providers. Supermicro’s motherboard, energy provide, and chassis design experience permits in-house growth and productiveness enhancements, bringing next-generation know-how from the cloud to the sting to prospects all over the world. Our merchandise are designed and manufactured within the United States, Asia, and the Netherlands, enhance world operations by way of productiveness and effectivity, and are designed for inexperienced computing to enhance TCO, cut back environmental influence, and enhance Server Building Block Solutions® Our distinctive, award-winning providing presents options in varied types and combos of processors, reminiscence, accelerators resembling GPUs, storage, networking, energy provides, and cooling strategies (air or liquid cooling). , you possibly can create the proper configuration on your wants, enhance your purposes and way more.

See also  The Combination of Two: Unveiling the Key to Chinese Characteristics

Supermicro, Server Building Block Solutions, and We Keep IT Green are logos and/or registered logos of Super Micro Computer, Inc.

Intel, the Intel brand, and different Intel symbols are logos of Intel Corporation or its subsidiaries.

All different manufacturers, names, and logos are the property of their respective house owners.

photograph – https://mma.prnasia.com/media2/2429019/Super_Micro_Plug_and_Play_Liquid_Cooled_AI_SuperCluster.jpg?p=medium600
brand –

(Japanese model: supplied by the consumer)

Company PR Newswire Asia Ltd.

PR Newswire
Founded in 1954, it’s the first American public relations information company on the earth. Our distribution community covers the entire world. As a subsidiary of Cision Ltd., Cision’s cloud-based communications merchandise, mixed with lots of the world’s largest channels, a multi-cultural content material distribution community and complete workflow instruments and platforms, energy the information of organizations all over the world.www.prnasia.com

You may also like

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More

Privacy & Cookies Policy