asitop-csv-logger 1.0.1

CSV logger for Apple Silicon using asitop

asitop-csv

Because sometimes you want your Apple Silicon metrics in spreadsheet form, not just pretty terminal colors.

A CSV-logging fork of the excellent asitop - a Python-based nvtop-inspired command line tool for Apple Silicon Macs. While the original asitop gives you beautiful real-time terminal output, this fork saves all that juicy performance data to CSV files for easy analysis with pandas, Excel, or your favorite data visualization tools.

What's Different?

Instead of just displaying metrics in your terminal (though it still does that), this fork logs everything to CSV files so you can:

• Create custom visualizations with matplotlib/seaborn
• Perform deep analysis with pandas
• Build dashboards in Jupyter notebooks
• Track performance trends over time
• Generate reports for system optimization

Perfect for developers, researchers, and data nerds who want to understand their Apple Silicon performance patterns beyond the moment.

Features

All the goodness of the original asitop, plus CSV logging:

Utilization Metrics (Now in CSV!)

• CPU utilization (E-cluster and P-cluster)
• GPU utilization and frequency
• ANE (Apple Neural Engine) utilization
• All timestamped for trend analysis

Memory Metrics (Spreadsheet Ready)

• RAM and swap usage over time
• Memory pressure indicators
• Perfect for identifying memory bottlenecks

Power Metrics (Chart-Friendly)

• CPU and GPU power consumption
• Peak power tracking
• Rolling averages for smooth trend lines
• Great for battery life optimization studies

Installation

Same as the original - you'll need pip and Python (which macOS already has):

pip install asitop-csv
# or if you're installing from source
pip install -e .

Usage

Basic CSV Logging

# Start logging to CSV (recommended - enter password upfront)
sudo asitop-csv

# Or let it prompt for password
asitop-csv

Advanced Options

asitop-csv [-h] [--interval INTERVAL] [--color COLOR] [--avg AVG] [--output OUTPUT] [--no-display]

optional arguments:
  -h, --help            show this help message and exit
  --interval INTERVAL   Display interval and sampling interval (seconds, default: 1)
  --color COLOR         Choose display color (0~8, default: 2)
  --avg AVG            Interval for averaged values (seconds, default: 30)
  --output OUTPUT      CSV output directory (default: ./asitop_logs/)
  --no-display        Disable terminal output, CSV logging only

Pro Tips

Long-term monitoring:

# Run in background, log every 5 seconds, no terminal spam
sudo asitop-csv --interval 5 --no-display --output ~/Desktop/my_mac_stats/

Benchmark session:

# High-frequency logging during testing
sudo asitop-csv --interval 0.5 --output ./benchmark_$(date +%Y%m%d_%H%M%S)/

CSV Output Format

Your data gets saved in timestamped CSV files:

asitop_logs/
├── asitop_20241216_143022.csv
├── asitop_20241216_150315.csv
└── ...

Each CSV contains columns like:

• timestamp - When the measurement was taken
• cpu_util_total - Overall CPU utilization %
• cpu_util_ecores - Efficiency cores utilization %
• cpu_util_pcores - Performance cores utilization %
• gpu_util - GPU utilization %
• cpu_power - CPU power consumption (W)
• gpu_power - GPU power consumption (W)
• memory_used - RAM usage (GB)
• memory_pressure - Memory pressure indicator
• And much more...

Quick Analysis with Pandas

This fork includes built-in pandas integration for easy visualization and analysis:

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
df = pd.read_csv("asitop_metrics.csv")
df['timestamp'] = pd.to_datetime(df['timestamp'])
sns.set(style="whitegrid")
metrics_to_plot = [
    "E-CPU Usage (%)", "P-CPU Usage (%)",
    "GPU Usage (%)", "RAM Used (GB)", "Swap Used (GB)",
    "CPU Power (W)", "GPU Power (W)", "Package Power (W)",
    "E-CPU Freq (MHz)", "P-CPU Freq (MHz)", "GPU Freq (MHz)"
]
for col in metrics_to_plot:
    plt.figure(figsize=(12, 4))
    sns.lineplot(data=df, x="timestamp", y=col)
    plt.title(col + " Over Time")
    plt.xlabel("Timestamp")
    plt.ylabel(col)
    plt.xticks(rotation=45)
    plt.tight_layout()
    plt.savefig(f"{col.replace(' ', '_').replace('(', '').replace(')', '')}.png")
    plt.close()
    
print(" graphs saved as ")

The tool automatically generates several common visualizations using pandas and matplotlib, including:

• CPU and GPU utilization trends
• Power consumption patterns
• Memory usage over time
• Temperature correlations
• Performance efficiency metrics

How It Works

Same underlying tech as the original asitop:

• powermetrics for CPU/GPU/ANE metrics (requires sudo)
• psutil for memory and swap stats
• sysctl for system information
• system_profiler for hardware details

Plus: Custom CSV writer that timestamps and saves all metrics without impacting performance.

Why This Fork?

The original asitop is fantastic for real-time monitoring, but sometimes you need the data for:

• Performance regression analysis
• Battery optimization studies
• Workload characterization
• Custom dashboard creation
• Academic research
• Debugging performance issues

This fork gives you the best of both worlds - live monitoring AND historical data analysis.

Requirements

• Apple Silicon Mac (M1, M1 Pro, M1 Max, M1 Ultra, M2, M2 Pro, M2 Max, etc.)
• macOS Monterey or later
• Python 3.6+
• sudo access (because powermetrics needs it)

Contributing

Found a bug? Want to add a feature? PRs welcome! This is a community-driven fork focused on making Apple Silicon performance data more accessible for analysis.

License

Same as the original asitop - because we're standing on the shoulders of giants.

Credits

Huge thanks to @tlkh for creating the original asitop. This fork simply adds CSV logging capabilities to their excellent work.

Disclaimers

Just like the original: "I did this randomly don't blame me if it fried your new MacBook or something."

But seriously, this tool only reads performance metrics - it doesn't change any system settings. The biggest risk is filling up your disk with CSV files if you forget to stop it.

---

Happy monitoring! May your CPU temps be low and your GPU utilization be high.