NVIDIA T4 Powers Next Generation of Virtual Workstations

Turing architecture-based T4 is the most flexible GPU for virtualization workloads, delivering on performance and user density.
by Emily Apsey

The NVIDIA Tesla T4 GPU now supports virtualized workloads with NVIDIA virtual GPU (vGPU) software.

The software, including NVIDIA GRID Virtual PC (GRID vPC) and NVIDIA Quadro Virtual Data Center Workstation (Quadro vDWS), provides virtual machines with the same breakthrough performance and versatility that the T4 offers to a physical environment. And it does so using the same NVIDIA graphics drivers that are deployed on non-virtualized systems.

NVIDIA launched T4 at GTC Japan as an AI data center platform for bare-metal servers. It’s designed to meet the needs of public and private cloud environments as their scalability requirements grow. It has seen rapid adoption, including its recent release on the Google Cloud Platform.

The Tesla T4 is the most universal GPU to date — capable of running any workload to drive greater data center efficiency. In a bare-metal environment, T4 accelerates diverse workloads, including deep learning training and inferencing as well as graphics. Support for virtual desktops with GRID vPC and Quadro vDWS software is the next level of workflow acceleration.

Roughly the size of a cell phone, the T4 has a low-profile, single-slot form factor. It draws a maximum of 70W power, so it requires no supplemental power connector.

Specifications for NVIDIA Tesla GPUs for virtualization workloads.

Its highly efficient design allows NVIDIA vGPU customers to reduce their operating costs considerably and offers the flexibility to scale their vGPU deployments by installing additional GPUs in a server. Two T4 GPUs can fit into the same space as a single Tesla M10 or M60 GPU, which could consume more than 3x the power.

The T4 is built on NVIDIA’s Turing architecture — the biggest architectural leap forward for GPUs in over a decade — enabling major advances in efficiency and performance.

Some of the key features provided by the Turing architecture include Tensor Cores for acceleration of deep learning inference workflows and new RT Cores for real-time ray tracing acceleration and batch rendering.

It’s also the first GPU architecture to support GDDR6 memory, which provides improved performance and power efficiency versus the previous generation GDDR5.

The Tesla T4 is an RTX-capable GPU, supporting the enhancements of the RTX platform, including:

  • Real-time ray-tracing performance
  • Accelerated batch rendering for faster time to market
  • AI-enhanced denoising to speed creative workflows
  • Photorealistic design with accurate shadows, reflections and refractions

The T4 is well-suited for a wide range of data center workloads, including:

  • Virtual desktops for knowledge workers using modern productivity applications
  • Virtual workstations for scientists, engineers and creative professionals
  • Deep learning inferencing and training

Read the full T4 Technical Brief for virtualization.

Check out what Cisco is saying about the Tesla T4, or find an NVIDIA vGPU partner to get started.

Learn more about GPU virtualization at GTC in Silicon Valley, March 17-21.