This release now provides 120Hz VSync support at resolutions up to 4K, which improves both VR and AR capabilities, and also facilitates collaborate real time rendering of high level graphics from employees distributed in home locations.
Today, NVIDIA is announcing the July 2020 version of their vGPU virtual GPU software. This version features enhancements which gives creatives and engineers the ability to collaborate from home-based environments. It also provides support for more workloads and use cases, and new efficiencies for IT admins.
“The ability of virtualization to address business continuity challenges has always been there, and people were doing remote work before,” said Anne Hecht, Senior, Director Product Marketing, GRID at NVIDIA. “This was already on the roadmap, but COVID made it more of a priority.”
The big technical enhancement to facilitate much of this is the addition of 120Hz VSync support at resolutions up to 4K, which was achieved through collaborative work with VMware and Ericcson. That improves the ability of NVIDIA CloudXR technology using the vGPU software to deliver virtual reality [VR] and augmented reality [AR]. It also facilitates doing high-level workloads like graphic rendering from home, because it providers a level of smoothness in the immersive experience on untethered devices that is basically the same as in tethered configurations.
“Adding this 120Hz VSync support and improving the remote immersive user experience opens up new real-time collaboration use cases for home-based workers,” Hecht said. “For traditional knowledge workers, it was fine before, but creative workers and engineers with higher requirements need this enhanced capability. Now, if one person viewing the model makes a change, it shows up on everyone else’s machine as well, in real time. This is incredibly powerful, to allow everyone to view a digital design simultaneously with real-time rendering.”
The improved VSync capability now lets the reference design support up to two virtual workstations on an NVIDIA Quadro RTX GPU, by being able to have two workstreams flowing through a virtual machine.
“This lets you put two users running complex on a single vGPU because of the enhanced VSync capacity,” Hecht indicated.
Hecht also noted that while the new capabilities directly benefit specialist users with high level workstation requirements, some customers see practical applications of this that are broader than that.
“One of our biggest deals is with a huge retailer who is positioning thousands of virtual workstations with Grid VPC to their employees, at least 6000 of them, with everyone working from home,” she said. “These employees are knowledge workers, but thus retailer decided for the vGPU. That’s because, in addition to the increased performance, vGPUs do bring savings over the cost of the device’s lifetime. There are fewer help desk calls. The VMs don’t crash so users are more productive.” Hecht noted that IDC interviews of NVIDIA customers using GPU-accelerated virtual desktops found organizations with 500-1,000 users saw a 13% increase in productivity, and more than $1 million in annual savings.
“The vGPU flexibility here also means GPUs are always provisioned to the people who need them, unlike physical GPUs, where if one person isn’t using it, it’s not assigned to someone else,” Hecht added.
The July vGPU release also adds features to help both enterprise IT admins and cloud service providers streamline their management. It supports VMware’s Distributed Resource Scheduler’s support for GPU-enabled VMs in vSphere, particularly a new feature called “Assignable Hardware” in vSphere 7 which lets a VM be automatically “placed” on a host with exactly the right GPU and profile available before powering it on.
“Distributed Research Scheduler [DRS] is important for vSphere customers, and being able to find a host immediately is a huge time saving, which will greatly speed up deployment time of new VMs,” Hecht said. “We will be bringing more DRS functionality in.”
Other management enhancements include cross-branch support, where the host and guest vGPU software can be on different versions, and support for the latest release of VMware vRealize Operations, which will provide enterprise data centers running vSphere with improved operational efficiency by having the ability to manage vGPU powered VMs with the latest release of VMware vRealize Operations.
“We have a lot of cloud service providers and enterprises that use us, so these features are important,” Hecht said.
“If you need virtualization and vGPU, there isn’t another solution that’s supports our range of use cases up to VR, and provides the same degree of IT management and monitoring support,” Hecht concluded.
The NVIDIA vGPU July 2020 release should be available by mid-July.