This test study is not intended to recommend specific instance sizes or types. Rather, it shows that by adjusting hardware resources and observing the results, a system can be tuned to increase the amount of work staff can do while improving their experience and increasing return on investment. However, adding more hardware resources without understanding the impact may not deliver the expected results.
Therefore, every organization should perform their own testing to evaluate the right hardware that effectively balances cost and performance for them, such as determining how much GPU is needed to support their workflows. Infrastructure needs change regularly, and routine testing should be done to optimize infrastructure investments.
Properly resourced desktop clients are essential to delivering a positive user experience, increasing editing efficiency, and increasing overall return on investment on your infrastructure. Therefore, make hardware choices that strike the balance between mitigating infrastructure expense (the cost of more robust instances) and operational expense (the cost of staff’s time, business interruption, and opportunity cost). ArcGIS Pro desktops should be GPU-enabled and should be allocated sufficient CPU for the workload. Learn more about ArcGIS Pro virtualization and GPU hardware selection in the ArcGIS Architecture Center.