Announcing IBM Cloud instances for NVIDIA H100 Tensor Core GPUs
IBM Hybrid Cloud and Infrastructure
IBM Hybrid Cloud and Infrastructure has the tools to make AI real. It’s your cloud. We protect it. You control it.
Written by Suraksha Vidyarthi , Product Manager, IaaS, IBM Cloud
We are thrilled to announce the availability of NVIDIA H100 Tensor Core GPU instances on IBM Cloud that accelerates a variety of AI and HPC workloads. As customers scale their Generative AI implementations from pilot to production, accelerators that are optimal for each stage of the AI lifecycle are needed. On IBM Cloud, customers now get a portfolio of NVIDIA accelerators - H100, L40S and L4 - to choose from based on their use cases and price performance considerations.
The IBM Cloud instance for NVIDIA H100, gx3d-160x1792x8h100, has eight GPUs connected using fourth generation NVLink technology, paired with two Intel Sapphire Rapids CPUs, 61TB of instance storage on a single server NVIDIA HGX H100 system.? The table below provides details on significant performance improvement this offering provides over the existing NVIDIA HGX A100, gx2-80x1280x8a100, instance type.
Additionally, IBM has improved overall availability of the NVIDIA H100 GPUs by making these instances available in 9 multizone regions across Americas, APAC, Europe and Japan.
gx3d-160x1792x8h100 is optimized for AI model fine tuning and inferencing, and traditional HPC simulation workloads.
watsonx: https://www.ibm.com/watsonx
InstructLAB: https://research.ibm.com/blog/instruct-lab
领英推荐
HPC solutions: https://www.ibm.com/high-performance-computing
Data privacy, security and compliance
AI and HPC workloads face several security threats that?can lead to data leakage and data privacy violations. IBM Cloud applies multi-level security protocols to protect AI data and processes and offers enterprises built-in controls to establish infrastructure and data guardrails for AI workloads.
AI model governance
A wide variety of tools exists for AI governance—but too often, models are built without proper clarity, monitoring or cataloging. Operationalizing AI with confidence is not possible without end-to-end AI lifecycle tracking using automated processes. Built on the IBM? watsonx AI and data platform, watsonx.governance helps direct, manage and monitor organization’s AI activities. Before a model is put into production, it’s validated to assess business risks. Once the model goes live, it’s continuously monitored for fairness, quality and drift. Regulators and auditors can get access to documentation that provides explanations of the model’s behavior and predictions.
Deployment automation
Securely deploying AI-powered applications in production require multiple services that include AI lifecycle management solutions, serverless platform, storage, security and compliance monitoring solutions etc. Manually configuring them takes time and increases likelihood of errors. IBM Cloud automates the deployment of these services and help reduce the risks of configuration errors.
NVIDIA H100 instances on IBM Cloud are available in North America, Latin America, Europe, Japan and Australia MZRs.
Looking forward to seeing how IBM Cloud customers leverage this innovation to drive their AI initiatives! IBM Hybrid Cloud and Infrastructure
Sr. Delivery Manager -Talent Acquisition at Connectpro Management Consultants PVT. LTD
5 个月Love it
Python Machine Learning | Full-Stack Developer | AI | Data Science | Research
5 个月Exciting news! ?? The availability of NVIDIA H100 Tensor Core GPU Instances on IBM Cloud is a significant step forward for enterprises looking to harness the power of AI. The combination of NVIDIA’s accelerated computing platform with IBM Cloud’s infrastructure will no doubt enable faster AI training and fine-tuning at scale, while potentially reducing costs. I’m looking forward to seeing how this partnership will unlock new possibilities for #GenAI in enterprise solutions. ??