WekaIO, the data platform software provider for AI, has announced its certification for an NVIDIA DGX BasePOD reference architecture. The architecture, built on NVIDIA DGX H100 systems and the WEKA Data Platform, offers massive data storage throughput and optimization for AI workloads. This certification marks a significant milestone for WekaIO and sets the stage for its journey towards DGX SuperPOD certification.
WekaIO Empowers AI Workloads with Certified NVIDIA DGX BasePOD Architecture
WekaIO has successfully received certification for an NVIDIA DGX BasePOD reference architecture, bringing enhanced performance and efficiency to AI workloads. The architecture is built on NVIDIA DGX H100 systems and powered by the WEKA Data Platform. With a rack-dense design, this solution delivers exceptional data storage throughput starting at 600GB/s and 22M IOPs in just 8 rack units, enabling organizations to optimize their DGX H100 systems.
The WEKA Data Platform serves as a critical data infrastructure foundation, transforming static data storage silos into dynamic data pipelines. This software-based architecture supports performance-intensive workloads like generative AI model training and inference at scale. It seamlessly fuels data-starved GPUs, empowering AI workloads across on-premises, cloud, edge, and hybrid environments.
The DGX BasePOD architecture provides the flexibility to scale resources according to evolving computational needs. By integrating NVIDIA H100 Tensor Core GPUs and NVIDIA InfiniBand networking technologies, this solution offers enhanced performance for diverse AI workloads, enabling faster model training and deployment. The DGX BasePOD is designed to prioritize agility, efficiency, and performance, making it an ideal choice for organizations seeking to optimize their data infrastructure investments while driving AI innovation.
The WekaIO and NVIDIA DGX BasePOD reference architecture seamlessly delivers the performance required by enterprise customers to accelerate AI adoption and achieve faster insights, discoveries, and outcomes. With 10x the bandwidth and 6x more IOPs than the previous configuration, this architecture sets a new standard for extreme performance in demanding AI workloads.
Key benefits of the WEKA with NVIDIA DGX BasePOD reference architecture include optimal efficiency, excellent linear scaling, and turnkey choice and flexibility. With its space- and energy-efficient footprint, it can support up to 16 DGX H100 systems and larger clusters of 32 or more systems.
This certification for the DGX BasePOD architecture marks a significant achievement for WekaIO and paves the way for the company’s DGX SuperPOD certification. By collaborating closely with NVIDIA and leveraging technologies like GPUDirect Storage (GDS), WekaIO continues to push the boundaries of AI infrastructure and deliver innovative deployment options for its customers. The journey towards DGX SuperPOD certification holds exciting possibilities, as WekaIO aims to provide even more advanced solutions for accelerating AI initiatives and delivering data-fueled insights.
The source of the article is from the blog mgz.com.tw