Instruction Trace Generation for Performance Analysis

[Ghliss]

Hello Pulpino community,

I'm working on performance analysis of code running on PULPino and need to generate detailed instruction traces with cycle timing.

Could you share how the PULPino development team typically generates and analyzes instruction traces for your own performance evaluation cycle accurately ?

Specifically:

1. What tools or methodologies do you use to extract cycle-accurate instruction traces during simulation?
2. Are there existing hooks in the debug infrastructure or simulation environment that support this type of tracing?
3. What would you recommend as the most straightforward approach to capture instruction execution on a cycle-by-cycle basis?

Thank you in advance.