NVIDIA DGX A100 is the foundational building block of AI infrastructure and part of the NVIDIA end-to-end data center solution created from over a decade of AI leadership by NVIDIA. As the universal system for all AI workloads, DGX A100 offers unprecedented compute density, performance and flexibility in the world’s first 5 petaFLOPS AI system.

Built on the revolutionary NVIDIA A100 Tensor Core GPU, DGX A100 unifies data center AI infrastructure, running training, inference, and analytics workloads simultaneously with ease. It integrates exclusive access to a global team of AI-fluent experts that offer prescriptive planning, deployment, and optimization expertise to help fast-track AI transformation.

During this webinar, attendees learned:

• The 3 key computational workloads of Artificial Intelligence and Machine Learning (AI/ML)
  • Training, inference, and analytics
  • Challenges in optimizing workloads across different resources
• The various AL/ML data types and how they are accelerated through Tensor Core processing
  • Tradeoffs of FP32 vs FP16 in traditional ML training
  • BFloat32 as a compromise to F16 limitations
  • New TF32 data type, performance gains, and how to use it
• How to access pre-trained models, containers, and helm charts via NVIDIA’s NGC to speed development time and time to solution