This is a script to visualize and compare CPU performance across multiple AI model workloads — LLaMA 2, LLaMA 3, and DeepSeek — using EMON data.
- Extracts key CPU performance metrics (e.g., utilization, frequency, CPI, license throttling levels).
- Compares metrics across LLaMA 2, LLaMA 3, and DeepSeek.
- Automatically adjusts:
- LLaMA models use system-wide metrics from
details system view - DeepSeek uses socket-specific metrics from
details socket view(specifically socket 0, since only one socket was active).
- LLaMA models use system-wide metrics from
- Generates clean, dot-and-line charts for each metric.
Required files: All 3 EMON data files
Required libraries:
pip install pandas matplotlib openpyxl