System monitor

This example turns Serial Studio into a live system dashboard, like a minimal htop, using the Process I/O driver (Pro) to launch a Python script and read its stdout directly as a telemetry stream.

Serial Studio spawns system-monitor.py (or the platform launcher), and the script continuously writes metrics to stdout at about 30 Hz. Serial Studio parses each line as a frame and drives plots, multiplots, bar graphs, and data grids in real time.

The Process I/O driver needs a Serial Studio Pro license. See serial-studio.com for details.

What it monitors

Frame format

The script emits two types of frames to stdout.

1. Header frame (emitted once at startup)

Static system information that populates the System Info data grid:

$OS=Darwin 25.3.0|HOSTNAME=myhost|CPU_MODEL=Apple M2 Pro|CPU_CORES=12|CPU_THREADS=12|RAM_TOTAL=17.2 GB|DISK_TOTAL=494.4 GB

2. Live frames (emitted at ~30 Hz)

Real-time resource usage, per-core CPU, and top processes:

$CPU_USAGE=9.4|RAM_USAGE=62.1|RAM_USED=8.50|DISK_USAGE=18.3|DISK_USED=12.3|CPU_TEMP=N/A|SWAP_USED=1.7|NET_SENT=0.124|NET_RECV=0.871|PROCESSES=682|CORE0=17.2|CORE1=16.0|...|CORE31=-1|PROC0=WindowServer (12.5% CPU)|PROC1=python3 (8.1% CPU)|...

All frames use $ as frameStart and \n as frameEnd.

Core sentinel value

The project always declares 32 core slots (CORE0 to CORE31). Slots beyond the machine's actual logical core count are emitted as -1. The multiplot widget treats -1 as out-of-range and hides those curves automatically.

How to run

Step 1: install dependencies

pip install psutil py-cpuinfo

• psutil. Cross-platform CPU, memory, disk, network, and process counters.
• py-cpuinfo. Reliable CPU brand string on all platforms (including Apple Silicon, where platform.processor() returns only "arm").

The platform launchers (system-monitor.sh / system-monitor.bat) auto-install both packages via pip if they're missing.

Step 2: configure Serial Studio

1. Open Serial Studio and load system-monitor.ssproj.
2. In the Setup panel, set Bus Type to Process.
3. Set Mode to Launch.
4. Set Executable to one of:
   • macOS/Linux: full path to system-monitor.sh.
   • Windows: full path to system-monitor.bat.
   • Or point directly to python3 and set Arguments to the full path of system-monitor.py.
5. Leave Working Directory empty (the script resolves its own path).
6. Click Connect.

Optional: run from the terminal

# macOS / Linux
./system-monitor.sh

# Windows
system-monitor.bat

# Direct Python (any platform)
python3 system-monitor.py --interval 0.033

Press Ctrl+C to stop.

Arguments:

How the frame parser works

The JavaScript parser inside system-monitor.ssproj uses a persistent value array declared at module scope. Keys absent from a frame keep their last value, so the static header fields (OS, hostname, CPU model, and so on) stay visible while live metric frames keep updating the rest.

// Module-scope globals, persist across parse() calls in QJSEngine
var _map  = { "OS": 0, "CPU_USAGE": 7, "CORE0": 17, ... };
var _vals = new Array(59).fill('');

function parse(frame) {
    var pairs = frame.split('|');
    for (var i = 0; i < pairs.length; i++) {
        var eq  = pairs[i].indexOf('=');
        var key = pairs[i].substring(0, eq).trim();
        var val = pairs[i].substring(eq + 1).trim();
        if (_map.hasOwnProperty(key))
            _vals[_map[key]] = val;
    }
    return _vals;
}

Dataset index N in the .ssproj maps to _vals[N - 1] (Serial Studio uses 1-based dataset indices, the JS array is 0-based).

Index layout

Architecture notes

• CPU sampling runs on a dedicated background thread using psutil.cpu_percent(interval=1.0, percpu=True). That's necessary on macOS, where interval=None always returns 0.0 until at least one blocking measurement has completed. The main loop reads the latest snapshot non-blocking via a lock.
• Slow metrics (disk, temperature, process list) are refreshed at 1 Hz regardless of frame rate, to avoid I/O overhead at 30 Hz.
• Fast metrics (RAM, network) are sampled every frame.

Dependencies

• Python 3.8 or later.
• psutil. pip install psutil.
• py-cpuinfo. pip install py-cpuinfo.

License

Copyright (C) 2020-2025 Alex Spataru SPDX-License-Identifier: GPL-3.0-only OR LicenseRef-SerialStudio-Commercial