ChipChat

I/O monitor showing meters for disks and network interfaces.

Installation

ChipChat % pip install -r requirements.txt
ChipChat % pip install -e .

Configuration

Create a config file at ~/.config/chipchat/config.yaml:

columns: 2
devices: { wlan0, nvme0n1 }

Meter Configuration

Each device can have 1-3 meters. Options vary by device type:

Disk meters:

meters:
  - utilization: { color: blue}        # %-util in blue
  - iops: { max: auto }                # rd+wr IOPS as a % of all-time high
  - bandwidth: { label: bw, max: 12GB} # custom label, explicit 12 GB/s max
  - blank                              # empty row for spacing

Network meters:

meters:
  - pps {color:{read: red,write: green}} # Packets per second in Christmas colors.
  - bandwidth { halflife: 5m }           # Bandwidth with exponentially-decaying auto-scaling.
  - blank                                # Empty row for spacing

Text Configuration

Disk text:

text:
  - name                                    # Name of device
  - usage: { thresholds: [50%, 80%, 99%] }  # Usage rate of drive; thresholds apply different styles
                                            # Renders IFF disk.mount_points[] is set
  - temp: { downsample: 10 }                # Drive temperature; updated every 10th refresh
                                            # Note: requires password-less sudo for SATA drives

Network text:

text:
  - name                                    # Name of device
  - ssid: { style: "cyan" }                 # Network SSID w. custom style
  - signal:                                 # Signal-strength w. override thresholds & styles
      thresholds:
        - {val: -50, color: blue}
        - {val: -80, color: purple}
  - ip: { style: "green" }                  # IPv4 address w. custom style

Encoding Units

Scale suffices must be uppercase (K/M/G/T). Data quantities can be expressed in bytes (B) or bits (b).

Format	Meaning
12GB	12 gigabytes (per second)
300MB	300 megabytes (per second)
1Gb	1 gigabit (per second)
auto	Auto-scale based on observed peak

Temperatures can be expressed in Celsius (C), Fahrenheit (F) or percentage (%) of registered critical drive temp (NVMe only).

Auto-Scaling

All meters defaut to { max: auto }, which means auto-scaling is enabled. This means the maximum recorded value (i.e., total bytes read + written over the sampling period for bandwidth meters) since the chipchat process was launched is tracked and used as the denominator for the meter.

This design means there is a "warmup" period where meters render high values shortly after startup. The recorded peak values are displayed on exit. If desired, users can copy these values into their config as explicit maximums; this will eliminate the warmup period on subsequent launches.

Decaying Exponentials

If typical usage is far below the recorded all-time high, then meters may not render much or any signal. One way to avoid this is to specify a lower explicit maximum for the meter. Another way is to enable the decaying-exponential feature within auto-scaling mode.

If halflife is set for an auto-scaling meter, then the denominator used to render the meter will constantly decay using an exponential pattern. For example, with halflife: 10m a meter that records 1200 MB/s spike at the top of the hour will use 1200 MB/s as the denominator for the very next refresh, then a sightly smaller denominator for the next refresh, and so on. By ten minutes past the hour, the denominator will be 600 MB/s (assuming traffic has decreased and the denominator is never exceeded). By twenty minutes past, it'll fall to 300 MB/s.

By default, network utilization meters use this feature. The default net utilization feature is effectively implemented with the following configuration:

- bandwidth: { max: auto, halflife: 60s, label: util, color: yellow}

Acknowledgements

ChipChat was primarily authored by Claude Opus-4.5, with web search and extended thinking modes enabled.