Image courtesy of Aaron Koelker
Geographic information system (GIS) professionals who use Esri’s ArcGIS Pro know how taxing the application can be on workstation resources. In today’s altered work environment, both budget and remote work make it hard for organizations to provide capable workstations to all of their employees who use ArcGIS Pro for complex geospatial workflows. Through their partnership with NVIDIA, Esri has addressed this issue by delivering ArcGIS Pro on GPU-enabled, virtual workstations that can run on-premises or in the cloud. As a result, GIS users can now host their virtualized ArcGIS Pro environments on Oracle Cloud Infrastructure’s (OCI’s) highly performant infrastructure and run resource-intensive workflows from their home or office.
It’s also worth mentioning that because of a shift in development efforts, Esri is recommending that users transition their projects and workflows from ArcMap, Esri's legacy desktop application, to ArcGIS Pro. This shift means that more GIS users will require higher-end workstations that can wholly take advantage of the geoprocessing capabilities available with ArcGIS Pro.
Esri continues to enhance ArcGIS Pro’s 3D visualization and develop new spatial-analysis tools that use the power of GPUs. GPUs on OCI compute instances can handle diverse workloads in ArcGIS Pro and are sufficient for the lightest to the heaviest GIS users. A GPU can accelerate several functions of ArcGIS Pro, such as 3D map rendering, spatial analysis, and running trained machine learning models. Examples of some of the spatial-analysis tools that can use a GPU are Slope, Aspect, and Viewshed2. These tools are designed to use CUDA, a parallel processing technology developed by NVIDIA, and they have been extensively tested by Esri on virtual instances with dedicated GPUs. Benchmarks showed substantial improvements to 3D map rendering and processing time of these spatial analysis tools.
Another area in which you can realize the benefits of a GPU in ArcGIS Pro is with the Deep Learning toolset provided with the Image Analyst extension. ArcGIS Pro lets you integrate external machine learning frameworks such as TensorFlow, PyTorch, Keras, and others to train deep learning models. You can then use the trained models with ArcGIS Pro’s inference geoprocessing tools, such as the Detect Objects Using Deep Learning tool. This tool runs a trained deep learning model on an input raster to produce a feature class that contains the objects that it finds. It’s also proven to deliver significantly faster compute times with a dedicated GPU.
OCI customers can use GPU-powered virtual machines with the NVIDIA Quadro Virtual Workstation (Quadro vWS) image from Oracle Cloud Marketplace to deploy a virtual desktop that can deliver robust workstation performance to employees anywhere in the world. The Quadro vWS image uses virtual GPU technology to deliver highly effective, graphics-enhanced, virtualized workstations. This image is prepackaged with all the necessary Quadro graphics drivers so that all you have to do is install your software. The image paired with the raw performance of OCI GPU instances is comparable to costly sophisticated workstations. Currently, you can deploy the Quadro vWS image on OCI’s VM.GPU3.x line of VM shapes, which are equipped with one, two, or four NVIDIA V100 Tensor Core GPUs based on Volta architecture. Stay tuned for the upcoming availability of the newest-generation GPUs with Nvidia A100 Tensor Core GPUs on OCI GPU VM shapes.
First, you need to meet the following prerequisites:
Have a valid Oracle Cloud Infrastructure account and subscription. If you’re not already an OCI customer, try it out by signing up for a free 30-day trial with US$300 in credits.
Configure the security list on the network subnet to allow traffic in through port 3389 for RDP access.
Then, you can get started with these three steps:
From Oracle Cloud Marketplace, deploy the NVIDIA Quadro vWS image on a VM GPU shape, and then follow the wizard to launch the instance in your tenancy.
Figure 1: Quadro vWS on Oracle Cloud Marketplace
Note: If the necessary GPU shapes aren’t available, you might be trying to deploy in a region that doesn’t support GPU shapes or you might have to request a service limit increase on your tenancy.
Connect to the instance using a remote desktop client and install ArcGIS Pro from your MyEsri subscription.
Figure 2: Installing ArcGIS Pro
Figure 3: Mapping in ArcGIS Pro
That’s it! Now you can virtually perform your ArcGIS Pro workflows from any location on just about any device.