NVIDIA VIRTUAL GPUs MULTIPLIED

Supercharge Virtual GPU Performance

THE POWER OF MULTIPLE vGPUs

Multiple NVIDIA virtual GPUs (vGPUs) can now be deployed in a single virtual machine (VM) to scale application performance and dramatically speed up production workflows.

HOW IT WORKS

NVIDIA vGPU software enables every virtual desktop infrastructure (VDI) user to harness the power of an NVIDIA GPU.

GPU Sharing

For lighter workloads, multiple virtual machines (VMs) can share GPU resources with NVIDIA virtual GPU software.

GPU Aggregation

For larger models and increasingly demanding workflows, NVIDIA Quadro® Virtual Desktop Workstation (vDWS) and NVIDIA Virtual Compute Server (vCS) software allow multiple GPUs to be assigned to a single VM.

Peer-to-Peer Computing

NVIDIA® NVLink enables a high-speed, direct GPU-to-GPU interconnect that provides higher bandwidth for multi-GPU system configurations than traditional PCIe-based solutions.

Harness the Power to Work Faster

Train Deep Learning Models Faster with Multiple GPUs

A deep learning training workload running TensorFlow ResNet-50 with mixed precision can run up to 50 times faster with multiple NVIDIA V100 GPUs and vCS software than a server with only CPUs. Additionally, running this workload in a hypervisor-based virtual environment using vCS performs almost as well as running the same workload in a bare-metal environment.

Deep Learning Training Performance

Up to 50X faster with vCS & V100

Speed Time To Market and Work with Larger, More Complex Models

Engineering simulations run almost 7 times faster, more smoothly and securely when powered by multiple NVIDIA GPUs with Quadro vDWS, versus CPUs only. In some cases, they can be run for much lower cost than a vCPU only solution. Multiple vGPUs enable exponentially faster processing with higher fidelity models in a VDI environment.

Ansys Fluent Computational Fluid Dynamics (CFD)

Tests were run on a server with 2X Intel Xeon Skylake (6148 2.4 GHz Turbo - 3.6 GHz), NVIDIA Quadro vDWS software, Tesla V100 GPUs with 32Q profile, Driver - 410.53, 256 GB RAM, Cent OS 7.4 64-bit. Benchmark Model: Water Jacket Model, Unsteady RANS, Internal Flow, Fluid- Water, size 4, 20 time steps

Abaqus/Standard 2018 Elastomeric Bearing Model

Tests were run on a server with 2X Intel Xeon Skylake CPUs (Xeon 6148 2.4 GHz 32-core), NVIDIA Quadro vDWS software, Tesla V100 GPUs with 32Q profile, Driver - 410.53, 256 GB vRAM, Cent OS 7.4 64-bit. Benchmark Model: ~450-550 TFLOPs, 5.9M DOF, Highly Nonlinear Static, Axisymmetric model with non-axisymmetric loading and twist, Direct Spare Solver (Model courtesy: SIMULIA)

Find the Time to Perfect Your Work

Render photorealistic scenes up to 4X faster, from just about anywhere, enabling designers to run multiple iterations in less time.

SOLIDWORKS Visualize (IRAY)

Tests were run on a server with 2X Intel Xeon Gold (6154 3.0 GHz) CPUs, 512 GB RAM, RHEL 7.5, NVIDIA Quadro vDWS software, Tesla V100-32Q, Driver - 410.39, 256 GB RAM, Windows 10 x64 RS3

Blender Cycles (CUDA) Render Time

Tests were run on a server with 2X Intel Xeon Gold (6154 3.0 GHz) CPUs, 512 GB RAM, RHEL 7.5, NVIDIA Quadro vDWS software, Tesla V100-32Q, Driver - 410.39, 256 GB RAM, Windows 10 x64 RS3

Simulation-driven product development takes engineering simulation to another level and the ability to scale multiple GPUs really kicks it up a notch. With the latest NVIDIA vGPU (Quadro vDWS), our users can work faster, more smoothly and more securely from just about anywhere – and in some cases, for significantly less cost than a vCPU only solution.

- Sumit Sathe, Lead Software Developer, ANSYS

 

The flexibility of the new multi-GPU feature available with NVIDIA Quadro vDWS opens up powerful new rendering workflows to SOLIDWORKS Visualize users. The near linear performance scaling means they can iterate on their designs at lightning speed on professional virtual workstations, allowing our customers to arrive at their best design in the shortest amount of time.

- Brian Hillner, SOLIDWORKS Product Portfolio Manager

Inspiration can strike anywhere. Autodesk Maya users are now able to utilize up to four NVIDIA Tesla GPUs in a single virtual machine, so the workstation can create and render faster photorealistic scenes regardless of when or where our customers are.

- Chris Vienneau, Senior Director, Media & Entertainment, Autodesk

LEARN HOW YOU CAN TAKE ADVANTAGE OF MULTI-vGPU

View product release notes and requirements.

INDUSTRY APPLICATIONS FOR MULTI-vGPU TECHNOLOGY

Seismic interpretation, simulation

Simulation, modeling, and design

Federal Government

Simulation and training

Rendering

Rendering, modeling & design

PURCHASE NVIDIA VIRTUAL GPUs