Forward-looking: As with many enterprise-focused technologies, the devil is in the details when it comes to exactly how and where virtual GPUs can function. Given the wide range of different server virtualization platforms and the graphics driver optimizations required for certain workstation applications, it can be challenging to get promising sounding technologies, like vGPUs, to work in all environments.
The new work habits that we’ve all adjusted to because of the pandemic have led many companies to take a fresh look at how they can provide computing resources to people working from home. In some cases, this is giving new life to technologies, such as VDI (virtual desktop infrastructure), that provide server-based computing sessions to remote desktops.
In addition, companies have also had to figure out how to provide remote access to workers with very specific, and very demanding technical requirements, such as architects, product designers, data scientists, media creators, and other people who typically use workstations in an office environment.
One critical technology for these new challenges is server-based virtual GPUs, or vGPUs for short. Nvidia has built datacenter optimized GPUs for many years, and several years back made them a shareable and manageable resource through the introduction of Virtual GPU software. The company’s latest July 2020 vGPU software release (one of two they typically do per year), adds several enhancements designed to make these server-based graphics chips function in a wider variety of software operating environments, with better compatibility across more applications, and managed in an easier way.
“With the Quadro Virtual Workstation software multiple people can work on CAD, architecture, or other highly-demanding applications with real-time rendering on regular PCs”
Given the wide range of server virtualization platforms and the driver optimizations required for certain workstation applications, it can be challenging to get vGPUs to work in all environments. To address these needs, the new release adds native support for virtualization on SUSE Linux Enterprise Server-based infrastructure, which is often used by data scientists, and offers additional management optimizations for VMware-based environments.
The new release also expands the capabilities of the different level GPU drivers that Nvidia provides, thereby increasing the range of applications it can support. Even details like different versions of drivers can make a difference in compatibility and performance. The latest release gives IT managers the flexibility to run different driver versions on the server and on a client device. This capability, called cross-branch support, is critically important for shared resources like vGPUs, because one application running on one device may need one version of a driver, and another application on another device may require another one.
Real-time collaboration across multiple applications is also supported in this new release. For VR-based applications, the new software, in conjunction with Nvidia’s CloudXR platform, can provide support for untethered mixed reality headsets with 4K resolutions at up to 120 Hz refresh rates over WiFi and 5G networks.
With the Quadro Virtual Workstation software—one of the several levels of drivers that Nvidia makes available through its vGPU software—multiple people can work on CAD, architecture, or other highly-demanding applications with real-time rendering on regular PCs. For designers, engineers, and others working from home, this capability can allow them to function as they normally would in a workstation-equipped office.
Interest in the ability to get remote access to these graphically demanding applications has been extremely high during the pandemic, which should be surprising to no one. This also aligns with results from a newly completed survey by TECHnalysis Research of over 600 US-based IT managers about the impact that COVID-19 has had on their IT strategies, priorities, and computing programs.
According to the study, virtual desktop infrastructure (VDI) usage grew over 11% in just a few months, from 48% of companies saying they used server-based computing models at the beginning of the year to 59% who said they are using them now. Not all of those instances of VDI use virtual GPUs, of course, but they do represent a significant and critical portion of them.
Ongoing flexibility has become the mantra by which IT organizations and workers are adapting to new work realities. As a result, technologies, such as vGPUs, that can enable flexibility are going to be a critical part of IT managers’ toolkits for some time to come.
Bob O’Donnell is the founder and chief analyst of TECHnalysis Research, LLC a technology consulting and market research firm. You can follow him on Twitter @bobodtech. This article was originally published on Tech.pinions.
Image credit: Jelena Zelen