At GTC today, NVIDIA unveiled a number of updates to its DGX portfolio in order to power new breakthroughs in enterprise AJE development.
NVIDIA DGX H100 systems are now available for order. These infrastructure building blocks support NVIDIA’s full-stack enterprise AJAI solutions.
With 32 petaflops of performance at FP8 precision, -NVIDIA DGX H100 delivers a leap in efficiency for enterprise AJAJAI development. It offers 3x lower total cost of ownership and 3. 5x more energy efficiency compared to the previous generation.
New NVIDIA Base Command software, which simplifies and speeds AI development, powers every DGX system — from single nodes to DGX SuperPODs.
Also unveiled was NVIDIA DGX BasePOD — the evolution of DGX POD — which usually makes enterprise data-center AI deployments simpler and faster for IT teams to acquire, deploy and manage.
Many of the world’s AI leaders are building technological breakthroughs — from self-driving cars to voice assistants — using NVIDIA DGX systems and software, and the pace associated with innovation is not slowing down.
New NVIDIA Base Command Features
-NVIDIA Base Command provides enterprise-grade orchestration and cluster management, and it now features a full software stack for maximizing AI developer productivity, THIS manageability plus workload performance.
The workflow management features of Base Command now include support for on-premises DGX SuperPOD environments, enabling businesses to gain centralized control of AJE development projects with simplified collaboration with regard to project teams, and integrated monitoring in addition to reporting dashboards.
Base Control works with the NVIDIA AI Enterprise software suite, which is now included with every DGX system. The -NVIDIA AI software program enables end-to-end AI development and deployment with supported AI and even data science tools, optimized frameworks together with pretrained models.
Additionally , this offers enterprise-workflow management and additionally MLOps integrations with DGX-Ready Software providers Domino Data Lab , Run. ai, Weights & Biases not to mention NVIDIA Inception member Rescale. It also includes libraries that optimize and accelerate compute, storage and network infrastructure — while ensuring maximized program uptime, security and reliability.
Brand new DGX BasePOD Reference Architecture
DGX BasePOD provides a reference architecture regarding DGX techniques that incorporates design best practices for integrating compute, networking, storage and also software.
Customers are already using NVIDIA DGX POD to power the development of a broad range of enterprise applications. DGX BasePOD builds on the success regarding DGX POD with new industry solutions targeting the biggest AI opportunities, including natural language processing, healthcare as well as life sciences, and fraud detection.
Delivered as fully integrated, ready-to-deploy offerings through the NVIDIA Partner Network , DGX BasePOD solutions range in size, from two to hundreds of DGX systems, along with certified high-performance storage through NVIDIA DGX storage technology partners including DDN, Dell, NetApp , Pure Storage , VAST Data and WEKA .
Leaders Power AJAI Breakthroughs With DGX Systems
Enterprises around the particular world choose NVIDIA DGX systems in order to power their most advanced AI workloads. Among the AJAJAI innovators developing mission-critical AI capabilities on DGX A100 systems:
- ML research and product lab Adept is creating an AJE teammate powered by a large language model prototyped upon NVIDIA DGX Foundry , and then scaled with -NVIDIA A100 GPUs and NVIDIA Megatron about Oracle Cloud Infrastructure.
- Hyundai Motor Group will be using the 40-node DGX SuperPOD to explore hyperscale AJAI workloads.
- Telecom company KT is developing a LLM together with around 40 billion parameters for a new variety involving Korean-language programs, including the GiGA Genie smart speaker, using the -NVIDIA NeMo Megatron framework, NVIDIA DGX SuperPOD and -NVIDIA Base Order software.
- The University of Wisconsin-Madison is quickly bringing AJAJAI to medical imaging devices using NVIDIA DGX methods with typically the Flywheel research platform plus the -NVIDIA Clara healthcare application framework. Using this NVIDIA Federated Learning Application Runtime Environment, or NVIDIA FLARE , in collaboration with other hospitals, the university is securely training AI models in DGX devices for medical imaging, annotation and classification.
Learn more about the AJE breakthroughs run by -NVIDIA DGX programs by watching NVIDIA founder and CEO Jensen Huang’s GTC keynote in replay. And join the GTC session, “ Designing Your AJAI Center connected with Excellence , ” using Charlie Boyle, vice president with DGX systems at NVIDIA.