NVIDIA vGPU
NVIDIA RTX Virtual Workstation (vWS) software delivers powerful virtual workstations from data centers to an endpoint device. The vWS software provides easy access to GIS, 3D Engineering CAD, and Computer-aided Engineering (CAE) applications in a VDI environment. The ease of accessing datasets enables greater collaboration among professionals in the GIS sector. From an IT standpoint, vWS provides centralized management of virtual workstations and exceptional monitoring features. The following figure shows an NVIDIA vGPU with a vWS license selected:
NVIDIA vGPU software creates virtual GPUs that are mapped to underlying physical GPUs. These virtual GPUs are assigned to each virtual workstation. NVIDIA vWS licenses enable the powerful graphics driver for professional 3D applications. The driver, which is installed on each virtual machine (VM) host, enables graphics commands from the VMs to pass to the GPU hardware. NVIDIA vWS comes with multiple profile size options that cater to different use cases and varying graphics requirements. You can select the correct GPU and vGPU profile (frame buffer) size for your professional graphics use cases. For GPU, vGPU profile, and density recommendations, see NVIDIA’s RTX Virtual Workstation Sizing Guide.
The vWS software supports applications with the highest resolution graphics and enables professionals to work with multiple 4K, 5K, and 8K displays. The software supports up to four 5K displays or two 8K displays, enabling professionals in the GIS sector to work on the details of graphics with ease.
At the time of testing, NVIDIA T4 GPUs were used to support the vWS CAD and GIS use case. NVIDIA now recommend their latest Ampere GPU technology, specifically the NVIDIA A40 GPU, for these workloads. Details and T4 upgrade options are shown in the following sections.
NVIDIA T4 GPU
The NVIDIA T4 GPU is a flexible GPU that serves various workloads. The NVIDIA T4 Tensor Core GPU is based on the Turing architecture and comes with 2,560 CUDA cores and a 16 GB DDR6 memory. The T4 GPU operates at 70 W, providing higher energy efficiency and lower operating costs than its predecessors. It has a single-slot PCIe form factor. You can configure up to six T4 GPUs in a single PowerEdge R750 server. The T4 GPU comes with an enhanced NVIDIA NVENC encoder that can provide higher compression and better image quality with H.264 and H.265 (HEVC) video codecs. The NVIDIA T4 NVENC encoder provides up to 25 percent bit rate savings for H.265 and up to 15 percent bit rate savings for H.264.
NVIDIA A40 GPU
The NVIDIA A40 is a flexible GPU recommended for light to high-end vWS user use cases. The A40 is the recommended upgrade path for RTX8000/6000 and T4 GPUs. It is based on the Ampere architecture and comes with 10,752 CUDA cores, 336 Tensor cores, 84 RT cores, and 48 GB of DDR6 memory. The A40 operates at 300 watts and has a dual slot form factor. You can configure up to two A40s in a single PowerEdge R750 server. The A40 comes with an enhanced NVIDIA NVENC encoder that can provide higher compression and better image quality with the H.264 and H.265 (HEVC) video codecs.
NVIDIA A16 GPU
The NVIDIA A16 is a flexible GPU recommended for entry level vWS and lightweight user use cases. It is the recommended upgrade path for T4 and M10 GPUs. The NVIDIA A16 GPU is based on Ampere architecture and comes with 4 x 1,280 CUDA cores, 4 x 40 Tensor cores, 4 x 10 RT cores, and 4 x 16 GB DDR6 memory. The A16 operates at 250 watts and has a full height, full length (FHFL) dual slot form factor. You can configure up to two A16s in a single PowerEdge R750 server. The A16 comes with an enhanced NVIDIA NVENC encoder that can provide higher compression and better image quality with the H.264 and H.265 (HEVC) video codecs.