Your Browser is Out of Date

Nytro.ai uses technology that works best in other browsers.
For a full experience use one of the browsers below
Design Guide—Virtualizing GPUs for AI with VMware and NVIDIA Based on Dell Infrastructure
Introduction Architectural Concepts Solution Architecture
Architecture overview
Recommended configurations
Design considerations
Software architecture
Network architecture
Validation and Machine Learning Performance Conclusion References

Architecture overview

Architecture overview

  • IT administrators who want to run mixed workloads including neural network training, inference, or model development using VMware vSphere with NVIDIA GPUs on PowerEdge servers will find the information in this design guide beneficial. The following figure shows how PowerEdge servers, each with two NVIDIA GPUs and a ConnectX network adapter, can be implemented as part of a VMware vSphere cluster to support virtual GPUs for both VMs and containers:


    Figure 3.   PowerEdge servers with NVIDIA A100 GPUs and VMware vSphere

    Servers running VMware vSAN provide a storage repository for the VM and operating system requirements while the PowerScale resources provide data lake storage for analytical workloads.

    Using the MIG capability of the NVIDIA A100 GPUs, administrators can create vGPU profiles and assign them to the VMs. These VMs can run various workloads such as TensorFlow for training, Jupyter notebooks for interactive model development, and TensorRT for inference.