Virtual workstation providing multi-GPU performance
Significantly improve the work speed of engineers and designers
On the 12th, NVIDIA announced the Quadro Virtual Data Center Workstation (Quadro vDWS), a virtual workstation.

Quadro vDWS is designed to accelerate demanding graphics and handle workflows. With Quadro vDWS's multi-GPU performance, professionals can work remotely from any device, while designs and intellectual property are securely protected in the data center.

Creative professionals working on remote virtual workstations can render realistic visualizations 94% faster by using two Tesla V100 Tensor Core GPUs instead of a single Tesla V100. Additionally, engineers and designers can complete simulations approximately seven times faster than CPU-only systems using two Tesla V100 GPUs.

The latest version of NVIDIA Virtual GPU software ensures reliability and ease of management through features such as real-time migration. The new features provided by the latest NVIDIA vGPU version from October 2018 are as follows.

NVIDIA Quadro vDWS enables the execution of multi-GPU workloads. It dramatically improves virtual GPU performance by integrating the performance of up to four NVIDIA Tesla GPUs in a single virtual machine (VM) for graphics and computationally intensive rendering, simulation, and design workflows. The latest NVIDIA virtual GPU products aggregate multiple Tesla GPUs and support GPU sharing across multiple VMs. Red Hat becomes the first virtualization platform to include this feature in Red Hat Enterprise Linux 7.5 and Red Hat Virtualization 4.2 KVM versions.

Real-time migration is possible via VMware vMotion. Organizational IT departments can migrate NVIDIA GPU-accelerated VMs in real time without impacting users or causing scheduled downtime. This not only saves time and resources but also allows them to focus on driving more strategic projects or business innovation. This new feature is currently supported in VMware vMotion and Quadro vDWS, GRID vPC, and GRID vApp software products using vSphere 6.7 u1.

It supports the NVIDIA Tesla T4 GPU. It supports double the frame buffer in the same low-profile and single-slot form factor as the previous generation Tesla P4. When combined with multi-GPU technology, the new 70W Tesla T4 can support much more demanding workflows in virtual desktop infrastructure environments, including advanced rendering, simulation, and design.

▲ NVIDIA GPU Cloud supports AI workloads on VMs. NVIDIA GPU Cloud (NGC) assists AI researchers in their research activities by providing GPU-accelerated deep learning containers for TensorFlow, PyTorch, MXNet, TensorRT, and others. NGC's ready-to-run deep learning containers are tested on the latest version of Quadro vDWS. Pre-integrated GPU-accelerated containers include the NVIDIA CUDA toolkit, NVIDIA deep learning libraries, and an operating system (OS).