ArcGIS systems that are business and/or mission-critical have requirements that generally include:
While there are many design considerations that contribute to the achievement of these requirements, this test study focuses specifically on the impact of CPUs and GPU-enabled client hardware on the performance and user experience of ArcGIS Pro utility network information management editing workflows.
A GPU, or Graphics Processing Unit, is a specialized hardware component that can improve processing efficiency for many compute-intensive tasks. For ArcGIS Pro, using a configuration that is not GPU-enabled can result in CPU emulation of GPU capabilities, which can cause suboptimal performance for workflows that depend on graphics processing.
Most major cloud providers provide access to GPUs through various virtual machine (VM) offerings. However, it is important to check the compatibility matrix provided by the virtualization software vendor and Esri.
Learn more about GPU hardware selection.
A CPU, or Central Processing Unit, is a server’s core computational unit. The amount and quality of CPU resources required is dependent on specific workloads. Usage patterns collected through appropriate telemetry capture and monitoring practices can help identify bottlenecks and determine whether acceptable utilization thresholds are exceeded. This might signal that an increase in CPU allocation is needed.
When operating ArcGIS in the cloud, like AWS, Azure, and GCP, it is important to understand the ratio of virtual CPU (vCPU) to physical CPU when making hardware choices so that system components may be assigned appropriate resources. There is a 2:1 ratio of vCPU:CPU for all instances used in this test study, but some virtualization options may have different ratios, such as 1:1.