Phocus Resource Monitor for Apple Silicon

A tool that monitors Hasselblad Phocus 4.x resource usage on Apple Silicon Macs, generating annotated graphs showing memory, CPU, GPU, and Neural Engine activity.

What This Tool Does

This tool watches Phocus in the background and records:

• Memory usage (Phocus-specific) — How much RAM Phocus is consuming
• CPU usage (Phocus-specific) — How hard Phocus is working your processor
• GPU utilization (system-wide) — Graphics processor activity percentage
• GPU power draw (system-wide) — How much power the GPU is consuming
• Neural Engine power (system-wide) — Apple's ML hardware activity (confirms HNNR uses the ANE!)
• Swap usage (system-wide) — How much disk is being used as overflow memory

When you stop monitoring, it generates a publication-ready graph with all this data, plus a CSV file for further analysis.

Example: Monitoring HNNR (Hasselblad Neural Network Noise Reduction) on an M4 Pro MacBook. Note the Neural Engine activity in the bottom panel — this confirms HNNR is using Apple's dedicated ML hardware.

Why "System-Wide" for Some Metrics?

macOS doesn't expose per-application GPU or Neural Engine usage. Memory and CPU are definitely Phocus-specific, but GPU and ANE reflect your whole system. For best results, close other apps during testing.

---

Option 1: Using the App (Recommended)

The easiest way to run Phocus Monitor is with the launcher app, which provides a graphical interface.

First-Time Setup

Before using the app, you need to set up the monitoring script:

1. Download this repository (click the green "Code" button → "Download ZIP", then extract it, or use git clone)

2. Open Terminal (press Cmd + Space, type "Terminal", press Enter)

3. Navigate to the folder and set up Python:

   cd ~/Downloads/phocus-monitor
   python3 -m venv .venv
   .venv/bin/pip install -r requirements.txt

   What these commands do:

   • cd ~/Downloads/phocus-monitor — navigates to the folder you downloaded
   • python3 -m venv .venv — creates a "virtual environment," an isolated space for this project's Python packages that won't affect anything else on your system
   • .venv/bin/pip install -r requirements.txt — installs two packages into the virtual environment:
     • psutil — reads process information (memory, CPU usage) from macOS
     • matplotlib — generates the graphs

That's it! The app will find this installation automatically.

Running the App

1. Open Phocus (the app monitors an already-running Phocus)

2. Double-click Phocus Monitor.app (located in app/dist/)

3. On first run, macOS will ask for permission to control Terminal. Click Allow:

   

   This is required because the app launches a Terminal window to run the monitoring script with administrator privileges. The app does not access your documents or data — it only sends commands to Terminal.

4. Follow the prompts:

   • Choose where to save the output files
   • Optionally enter a duration in seconds (or leave blank to stop manually)
   • Set the sampling interval (default 2.0 seconds — lower values give more detail but larger files)
   • Click OK to start

5. Enter your Mac password in the Terminal window that opens (required for reading GPU and Neural Engine metrics)

6. Add annotations by pressing Enter in the Terminal and typing what you're about to do (e.g., "HNNR start - 3 images")

7. Stop monitoring — If you set a duration, the script will stop automatically and generate the graph. Otherwise, press Control+C (not Command+C) in the Terminal window to stop and generate the graph.

---

Option 2: Using the Command Line

For users comfortable with Terminal, you can run the script directly.

Requirements

• macOS on Apple Silicon (M1, M2, M3, M4 series)
• Phocus 4.x installed
• Python 3.9+ (included on macOS)
• Administrator access (the script needs sudo to read hardware metrics)

Installation

1. Open Terminal and navigate to where you want to save the script:

   cd ~/Downloads

2. Clone the repository (or download and extract the ZIP):

   git clone https://github.com/kmichels/phocus-monitor.git
   cd phocus-monitor

3. Set up the Python environment:

   python3 -m venv .venv
   .venv/bin/pip install -r requirements.txt

Running the Script

1. Open Phocus (the script monitors an already-running Phocus)

2. Run the script with sudo:

   sudo .venv/bin/python3 monitor_phocus.py

3. Enter your Mac password when prompted. (You need sudo because reading GPU and Neural Engine data requires administrator privileges.)

4. The script will start monitoring. You'll see output like:

   ╔════════════════════════════════════════════════╗
   ║   Phocus Resource Monitor v2.5.1              ║
   ╚════════════════════════════════════════════════╝
   
     System: Apple M4 Pro
       CPU: 14 cores (10P + 4E)
       GPU: 20 cores
       ANE: 16-core Neural Engine
       RAM: 64 GB
   
     Phocus: v4.0.1
   
     Interval: 2.0s
     Output: phocus_monitor_20241210_143052.*
   
     Controls:
       Press Enter to add an annotation
       Press Ctrl+C to stop and generate graph
   

5. Add annotations by pressing Enter and typing what you're about to do:

   > HNNR start - 3 images
   

   These labels appear on your graph, helping you remember what was happening at each point.

6. Perform your Phocus operations — import images, apply HNNR, browse thumbnails, whatever you want to measure.

7. Stop monitoring. If you used --duration, the script stops automatically. Otherwise, press Control+C (not Command+C) to stop. Either way, the script will generate:

   • A PNG graph file (e.g., phocus_monitor_20241210_143052.png)
   • A CSV data file (e.g., phocus_monitor_20241210_143052.csv)

Command Line Options

sudo .venv/bin/python3 monitor_phocus.py [options]

Option	Default	Description
--duration SECONDS	unlimited	Stop automatically after this many seconds
--interval SECONDS	2.0	How often to sample (lower = more detail, larger files)
--output PATH	auto-generated	Output path: directory or full path (see examples)
--version	—	Show version and exit

Examples:

# Monitor for exactly 5 minutes
sudo .venv/bin/python3 monitor_phocus.py --duration 300

# Sample every half-second for detailed analysis
sudo .venv/bin/python3 monitor_phocus.py --interval 0.5

# Specify output filename (saves as hnnr_test_run1.png and .csv)
sudo .venv/bin/python3 monitor_phocus.py --output hnnr_test_run1

# Specify output directory (uses default timestamped name in that directory)
sudo .venv/bin/python3 monitor_phocus.py --output ~/Documents/phocus-tests/

# Full path with filename
sudo .venv/bin/python3 monitor_phocus.py --output ~/Documents/phocus-tests/my_test

Note: If you specify a directory that doesn't exist, the script will ask if you want to create it.

---

Understanding the Output

The generated graph has 5 panels:

1. Memory (GB) — Phocus RAM usage (blue) and system swap (orange, right axis)
2. GPU Active (%) — System-wide GPU utilization
3. CPU (%) — Phocus CPU usage (100% = 1 full core, 200% = 2 cores, etc.)
4. GPU Power (W) — System-wide GPU power consumption
5. ANE Power (W) — Neural Engine power (spikes during HNNR confirm ML hardware usage!)

Vertical dashed lines show your annotations.

The summary bar at the bottom shows averages and maximums for each metric.

---

Troubleshooting

"Phocus is not running"

Start Phocus before running the script or app. The monitor watches an already-running Phocus process.

"Permission denied" or no GPU/ANE data

Make sure you're running with sudo (command line) or entering your password when prompted (app).

"No module named psutil" or "No module named matplotlib"

The virtual environment isn't set up or the packages aren't installed. Run:

cd ~/Downloads/phocus-monitor
python3 -m venv .venv
.venv/bin/pip install -r requirements.txt

The app shows "Setup Required"

The app couldn't find the phocus-monitor installation. Make sure you've:

1. Downloaded/cloned the repository
2. Created the virtual environment with python3 -m venv .venv
3. Installed dependencies with .venv/bin/pip install -r requirements.txt

The app looks in these locations:

• ~/phocus-monitor
• ~/Downloads/phocus-monitor
• ~/Documents/phocus-monitor
• The folder containing the app

The graph looks weird or has missing data

• GPU/ANE all zeros? Make sure you're using sudo or entering your password
• Memory seems too high? Phocus caches images aggressively. This is normal — see our findings about memory accumulation during browsing.

---

Tips for Good Data

1. Close other apps — GPU and ANE are system-wide metrics, so other apps will add noise
2. Use annotations liberally — It's hard to remember what you were doing at timestamp 3:42
3. Let things settle — Wait a few seconds between operations so you can see clear boundaries in the graph
4. Longer sessions = clearer patterns — 3-5 minute sessions work well

---

Uninstalling

This tool doesn't install anything system-wide — everything is contained in the project folder. To completely remove it:

# Delete the entire project folder (includes venv and all dependencies)
rm -rf ~/Downloads/phocus-monitor

That's it. The virtual environment (.venv/) contains all the Python packages, so deleting the folder removes everything.

If you want to keep your output files:

# Keep only your CSV and PNG files
cd ~/Downloads/phocus-monitor
cp *.csv *.png ~/Desktop/  # or wherever you want them
cd ..
rm -rf phocus-monitor

---

Share Your Results!

We'd love to see what you discover! Post your monitoring graphs in our GitHub Discussions.

When sharing, include:

• Your graph (drag and drop the PNG into the post)
• What you were doing (browsing thumbnails, applying HNNR, exporting, etc.)
• Any issues (did Phocus crash? freeze? behave unexpectedly?)

Your data helps the community understand how Phocus performs across different Mac configurations and workflows.

---

Contributing

Found a bug? Have an idea? Open an issue or PR on GitHub.

This tool was developed as part of a documentation project for Hasselblad Phocus. Read more at tonalphoto.com.

---

License

MIT License — see LICENSE for details.