Nvidia is rolling out virtualization tools that take its Tesla GPU-accelerated servers and allow enterprises to deploy virtual workstations.
The company is launching its Nvidia Quadro Virtual Data Center Workstation Software, or Quadro vDWS. The software streams any workstation or high-performance computing application from the data center to a laptop or virtual workstation.
Nvidia's bet is that more enterprise workloads will need compute intensive architecture and GPUs.
Quadro vDWS will allow more compute-heavy applications to cover dispersed teams. Nvidia is aiming to take the approach behind virtual desktops and apply it to high-end technologies used for 3D rendering, virtual reality, simulations, and deep learning in industries such as engineering.
According to Nvidia, Quadro vDWS runs on the company's Pascal architecture and enables:
3D designs with up to 24GB of GPU memory for large models;
Double the graphics performance of Nvidia's previous GPU architecture;
Unified graphics and compute workflows for design simulations;
And performance enhancements for Linux virtual workstation users.
In addition, Nvidia updated its Grid virtual PC efforts. The company updated its Grid release with its latest Tesla GPUs. On the virtual desktop front, Nvidia's updates include:
Support for up to 24 instances of 1GB virtual desktop profiles with the Tesla P40.
vGPU visibility at the host and end-point levels with application monitoring tools.
A resource scheduler to help IT departments allocate resources to users, cut latency and improve the user experience.
Nvidia's virtual desktop and workstation enhancements are available now via server systems from the likes of Cisco, Citrix, Dell, HPE, IBM, Lenovo and VMware. Some features will launch Sept. 1.
Moore's Law is slowing down as emerging AI workloads are demanding more performance. But Nvidia says it's got a plan. At this year's GPU Technology Conference, it announced powerful hardware and software for deep learning.