Open the run in Arm Performix for deeper inspection

Validate the insight in Arm Performix

A chat response helps you triage a performance problem, but the underlying profile is the source of evidence. Open the same run ID in Arm Performix and inspect the views relevant to the claim:

• Flame graph to see where sampled CPU time accumulates across call paths
• Functions to confirm which functions account for the largest sample percentages
• Call Stack to distinguish a function’s own cost from the cost of its callees
• Source to correlate samples with source lines when source mapping is available
• Disassembly to inspect generated instructions and confirm claims about the compiled code
• Compare to evaluate equivalent before-and-after runs

Use the AI insight to focus your investigation, not replace it. For example, if the insight says a loop appears scalar, inspect its disassembly and compiler flags before changing the implementation. If Code Hotspots only shows that the loop is hot, run Instruction Mix or another appropriate recipe before concluding that scalar execution is the cause.

For a before-and-after comparison, keep the workload, target, input, thread count, CPU affinity, and collection settings consistent. A lower sample percentage doesn’t prove that a function became faster if total runtime or the surrounding workload changed.

What you’ve accomplished

You’ve now configured the Arm Performix MCP server, selected or created a Code Hotspots run, generated an AI insight, and checked its recommendation against the profile.

You can continue with the Arm Performix getting-started video to explore the graphical workflow.