Powering Mixed Workloads with NVIDIA Virtual GPU

TL;DR

NVIDIA virtual GPU technology allows for efficient workload management, from running CAD workloads to HPC solvers, while providing a seamless user experience.

Transcript

here you can see that we have two server nodes showing up on the left side of your screen each of these servers is equipped with nvidia tesla datacenter GPUs and NVIDIA virtual GPU software both are hosting virtual machines running various workloads the HPC solver has been powered off and the servers are running CAD workloads let's minimize the man... Read More

Key Insights

• 🏃 NVIDIA virtual GPU technology enables fluid running of CAD workloads and intensive applications in a virtualized environment, providing a native workstation-like experience.
• 👻 Dynamic workload management, including live migration, allows for efficient resource utilization and enables the prioritization of high-performance computing tasks.
• ⌛ GPU acceleration speeds up HPC solvers by five to ten times, reducing the time required for complex numerical analysis.
• ♻️ The ability to analyze and refine models in a 24/7 computational environment accelerates workflow and supports continuous improvement.

Questions & Answers

Q: How does NVIDIA virtual GPU technology enable fluid running of CAD workloads and intensive applications?

NVIDIA virtual GPU technology leverages GPUs to power the virtualized environment, providing the necessary computational resources for seamless performance and a native physical workstation-like experience.

Q: How is workload migration handled in a virtualized environment?

Workload migration is achieved through live migration, where a VM can be moved from one server node to another without disruptions or loss of productivity for the user. This ensures efficient resource utilization and enables dynamic workload management.

Q: How does GPU acceleration improve the performance of HPC solvers?

By utilizing the GPU for HPC computations, the performance of the solver is accelerated by five to ten times compared to traditional CPU-only computations. This significantly reduces the time required for running complex numerical analyses.

Q: How does NVIDIA virtual GPU technology support data center workflow acceleration?

With features like live migration, the data center can work continuously to accelerate the workflow. This allows engineers to analyze and refine models, benefiting from 24/7 computational resources and accelerated performance.

Summary & Key Takeaways

• The content showcases the use of NVIDIA virtual GPU technology to run various workloads on server nodes equipped with NVIDIA Tesla GPUs.

• CAD workloads and intensive applications like WebGL and Paraview can be run fluidly in a virtualized environment, providing a workstation-like experience.

• Workloads can be dynamically managed, allowing for migration of virtual machines (VMs) between server nodes to free up resources for high-performance computing (HPC) tasks.