Profiling for Neoverse with Streamline CLI Tools: Review

Log an issue

Fork and edit

Discuss on Discord

What you've learned

You should now know how to:

Describe Arm's top-down profiling methodology.
Use Streamline CLI tools to capture and analyze performance data from an application.

How do Streamline CLI tools generate reports?

Using three distinct processes, `sl-record`, `sl-analyze` and `sl-format.py`.

Using the Streamline GUI.

What should you do if you see functions with a high branch mispredict rate and significant branch MPKI number?

Nothing, this is not an important metric.

Reduce branching in the affected functions.

Try to improve the predictability of branches, or convert unpredictable branches into conditional select instructions.

What is the Frontend phase responsible for?

The execution of micro-ops by the processing pipelines inside the core.

Instruction fetch, decode, and dispatch.

The resolution of micro-ops that are architecturally complete.

Back