feat: DEP for AIPerf Visualization Feature and Plot Command

[ilana-n]

Add AIPerf Visualization Feature Design Plan

Proposes aiperf plot command to generate visualizations from profiling results.

Key features:

• Three visualization modes: static PNG (default), interactive HTML (offline), and live server dashboard
• Auto-detects multi-run comparison vs single-run analysis
• Supports all AIPerf metrics with configurable plot presets
• User configuration via ~/.aiperf/config.yaml

Includes:

• Complete design specification with requirements and architecture
• Configuration examples for custom plot definitions
• Visual mockups showing example outputs
• Design plan template and reference materials

This addresses the common need to visualize performance metrics, compare runs, and analyze trade-offs without external tools.

[lkomali]

Good job on detailing the project. Liked the idea of interactive plots.
I have some comments, open to discuss.

In general, my opinion is that analyze and plot should be decoupled.

• profile will benchmark and generate results. Users might want to visualize individual runs as well as compare across runs.
• analyze will sweep across different values for a parameter and provide results. Users running analyze may or may not want to visualize their results.
• plot should be able to visualize results coming both the above subcommands. Users should be able to visualize given a set of result files irrespective of the step before ran profile or analyze command.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

✨ Finishing touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[the-david-oy]

This is a great proposal.

[lkomali]

I'm halfway through it, but really good design.
Have a couple of small comments.
Will take another pass for the other half.

[lkomali]

LGTM

Great work @ilana-n

[nv-kmcgill53]

I think there is room for a three dimensional plot for the pareto curves and metric comparisons since these metrics can change for the different ISL/OSL counts. Being able to make effective 3D visualizations would be pushing the frontier here.

[ilana-n]

Yes, that would be great, thanks for the feedback! Will add this - it will likely will be additional work after basic functionality is finished.

[ilana-n]

This work is on the backlog for AIPerf and being tracked here:
https://linear.app/nvidia/issue/AIP-329/integrate-server-side-metrics-with-aiperf-runs
https://linear.app/nvidia/issue/AIP-274/add-kv-cache-performance-reporting