CHATSWORTH, Calif. – March 21, 2023 – AI and multi-cloud information control corporate DDN as of late introduced compatibility with the following era of NVIDIA DGX techniques, each and every with 8 NVIDIA H100 Tensor Core GPUs. NVIDIA DGX H100 supercomputers are the fourth era of NVIDIA’s purpose-built AI techniques and designed for essentially the most taxing coaching workloads, reminiscent of herbal language processing and deep studying recommender fashions. A lot of these workloads require massive information fashions and high-speed throughput to ship leap forward effects, and pairing NVIDIA DGX techniques with DDN’s A3I is a mixture designed for AI facilities international.
DDN AI400X2 garage equipment compatibility with DGX H100 techniques construct on DDN’s field-proven deployments of DGX A100-based DGX BasePOD reference architectures (RAs) and DGX SuperPOD techniques which were leveraged by way of shoppers for quite a lot of use instances. Introduced as a part of DDN’s A3I infrastructure resolution for AI deployments, shoppers can scale to reinforce higher workloads with more than one DGX techniques. DDN additionally helps the most recent NVIDIA Quantum-2 and Spectrum-4 400Gb/s networking applied sciences. Validated with NVIDIA QM9700 Quantum-2 InfiniBand and NVIDIA SN4700 Spectrum-4 400GbE switches, the techniques are beneficial by way of NVIDIA in the latest DGX BasePOD RA and DGX SuperPOD. With double the IO functions of the prior era, DGX H100 techniques additional necessitate the usage of excessive efficiency garage answers like DDN’s AI400X2.
“The call for for scalable AI infrastructure continues to develop, as enterprises notice the ability that AI delivers to become their trade,” stated Dr. James Coomer, senior vice chairman for merchandise at DDN. “We see increasingly organizations which are shifting from assessing AI to making use of AI to ship trade effects. Those organizations are on the lookout for shown infrastructure that integrates into their information heart in a easy and environment friendly method, which is precisely what NVIDIA DGX techniques with DDN garage delivers.”
Along with those on-premises deployment choices, DDN could also be pronouncing a partnership with Lambda to ship a scalable information resolution in line with NVIDIA DGX SuperPOD with over 31 DGX H100 techniques. Lambda intends to make use of the techniques to permit shoppers to order between two and 31 DGX circumstances sponsored by way of DDN’s parallel garage and the total 3200 Mbps GPU material. This hosted providing provides fast get entry to to GPU-based computing with no dedication to a big information heart deployment together with a easy aggressive pricing construction. Lambda selected DDN because the backend garage for this challenge as a result of DDN’s established monitor report of a success DGX SuperPOD deployments, in addition to the experience for garage at scale that DDN brings to the desk. Lambda can be promoting DGX BasePOD and DGX SuperPOD with DDN A3I garage for patrons having a look to determine on-site deployments.
“As organizations proceed to modernize round AI, they’re experiencing explosive call for round efficiency and information wishes,” shared David Corridor, head of excessive efficiency computing, Lambda. “To handle that want, Lambda, as a marketplace chief within the deep studying infrastructure area, is bringing NVIDIA DGX techniques with DDN A3I garage into our reserved cloud providing. This gives our shoppers with a full-service revel in coupled with industry-leading efficiency in a question of weeks quite than months.”
Be told extra about DDN’s new RAs, why environment friendly excessive efficiency parallel garage provides a vital benefit for AI workflows, and Lambda’s GPU cloud by way of staring at DDN’s on-demand video, Unharness Lightning-Rapid Garage for Unheard of AI Potency and Efficiency (Introduced by way of DDN) [S52439], at NVIDIA GTC, an international convention on AI and the metaverse, operating on-line via March 23.
Supply By way of https://insidehpc.com/2023/03/ddn-announces-compatibility-with-nvidia-dgx-h100-and-lambda-partnership/
