Three years ago, we released the first version of the ArcGIS Pro Performance Assessment Tool, affectionately known around here as “PAT”. We wanted to design a tool that would be an easy-to-use, self-contained, means for benchmarking ArcGIS Pro’s performance on different hardware.
We have been overwhelmed by the positive support from our user community. Almost weekly we hear from you that you’ve deployed PAT in-house on new hardware, and on newly-deployed virtual machines both on-premises and in the cloud. From lightweight laptops to ruggedized hardware, you’ve used this solution to measure how fast ArcGIS Pro performs various critical GIS workflows. In instances where you’ve encountered performance issues, you’ve often been able to troubleshoot and re-run with improved results. Some of you have been gracious to share with us your PAT results.
To continue in our quest to make PAT more useful, we are pleased to announce a new release, version 17. While the underlying tests and expected timings remain exactly the same, we have added some enhancements to the core execution mechanism. Specifically:
- In the results output (SUMMARY_AllLevels.csv):
- We now check for test execution errors in the log files after execution. The attribute “TestStatus” shows any errors in the output
- Some test machines have more than two GPU’s/graphics adapters. We now show details for a third, in “GPU3”
- From user feedback, We’re appending a new GUID to the TestResults folder name to make for easier merging of PAT runs on multiple machines launched by scheduled tasks
- We have bundled the latest version of the PerfTools Add-in, a critical component for PAT execution. This include several stability and compatibility enhancements
- The accompanying PAT-README.PDF now contains detailed updates to document these additions, and contains more detailed information on test results output
With ArcGIS Pro 3.3 now released, and the 2024 Esri User Conference on the way, download and try it for yourself today!
Commenting is not enabled for this article.