Skip to content

Configuring Measures

Jump to bottom
Kurou edited this page Jul 9, 2026 · 4 revisions

Configuring Measures

Every reading is one Rainmeter measure with Measure=Plugin and Plugin=NativeHardwareMonitor. What it reads is set by Category + Metric; which device by Device. All option names and values are case-insensitive and whitespace-trimmed.

[MeasureGpuTemp]
Measure=Plugin
Plugin=NativeHardwareMonitor
Category=GPU
Metric=Temperature
Device=0

Options

Category — required

Which subsystem. One of GPU, CPU, Memory, Network, Storage, Ping, Battery, System, Motherboard, Display. Short aliases (e.g. net, mb, monitor) also work. An unknown category makes the measure inert (it reads -2) and logs a note.

Metric — required

Which reading, within the category. Values are category-dependent — see the full list in Metrics Reference, which also lists the short aliases each metric accepts. An unknown metric for a valid category behaves like an unknown category (inert).

Device — default 0

Which instance. Meaning depends on the category:

Category Device means
GPU GPU index in enumeration order (0 = first detected)
CPU 0 = whole-CPU aggregate; 1..N = individual logical core. Topology/identity metrics live on 0
Memory 0 = RAM, 1 = Virtual (commit), 2 = Swap (pagefile)
Network Interface index (loopback/tunnel excluded)
Storage Drive-lettered volume, sorted by drive letter
Ping not used — use Host= instead
Battery not used — always a single battery (0)
System not used — single machine-wide device (0)
Motherboard not used — single device (0), all-string identity
Display Monitor index (0 = primary/first active monitor)

Host — Ping only

Hostname or IP to ping (e.g. 1.1.1.1, google.com). Replaces Device= for the Ping category. Each distinct host gets its own background poller.

PingInterval — Ping only, default 1000

Milliseconds between ICMP echoes for this host.

UpdateOverride — default 0 (follow Rainmeter)

Milliseconds. When set, this category is polled on a background thread at this rate, independent of the skin's Update=. Omit it to gather in step with Rainmeter's tick. Useful for a fast network/storage readout under a slow skin, or to throttle an expensive category.

Do not use UpdateRate — that's a reserved Rainmeter keyword. The option is named UpdateOverride precisely to avoid the collision.

UpdateOverride vs UpdateDivider — sampling vs display

These are two different knobs and they do not conflict:

  • UpdateOverride (plugin) sets how often the hardware is sampled.
  • UpdateDivider (built-in Rainmeter) sets how often the measure is read/displayed — it just changes how often the cached value is shown, never how often it's sampled.

Set both freely: the hardware is sampled every UpdateOverride ms, and the meter shows the latest cached value every Update × UpdateDivider. Rate metrics (network/storage speed, ping) divide by real elapsed time, so they stay correct at any combination.

The common mistake is reaching for the wrong one. A big UpdateDivider on a Download speed measure with no UpdateOverride samples only when the meter is read, so "speed" becomes an average over the whole divider window. Add UpdateOverride=1000 and the background thread samples every 1 s while the meter still refreshes slowly — a true 1-second speed shown at your chosen display rate.

UpdateOverride is per-category, not per-measure. All measures of one category (e.g. every Category=GPU measure) share one background poller running at the smallest UpdateOverride set among them. Setting it on one measure sets the sampling floor for the whole category. To make a single measure update slower on screen, use UpdateDivider on that measure — it's free and doesn't touch sampling.

Debug — default 0

Diagnostic logging to the Rainmeter log (About → Log), plus an optional metric dump.

Value Behaviour
0 Silent (plus always-on init/error events)
1 Log this measure's registration/reload
2 All registration/reload events + dump every metric for this measure's category+device to a file

At Debug=2 the plugin writes %TEMP%\NativeHardwareMonitor\<Category>_dev<N>.txt (overwritten each update, throttled to ~2 s) listing every metric for that category and device — numeric values (with unsupported / error annotations) and strings. This is the file to send back when helping verify hardware the author can't test locally (see Intel GPU — help verify).

The plugin tracks the max Debug across all measures, so one measure at 2 raises verbosity globally.

Reading the value in a meter

Rainmeter measures expose a numeric side and a string side; the plugin fills whichever the metric is.

  • Numeric metrics (Usage, Temperature, clocks, bytes…) → %1 shows the number. Bind these to any meter (String, Bar, Line, Histogram, Roundline…).
  • String metrics (Name, Vendor, VolumeLabel, ThrottleReasons, MemoryType…) → %1 shows the text, but only in a Meter=String. Bound to a numeric-only meter (Bar/Line/Histogram) a string metric reads -1.
; numeric — clocks come out in Hz, so autoscale
[MeasureCoreClock]
Measure=Plugin
Plugin=NativeHardwareMonitor
Category=GPU
Metric=CoreClock
[MeterClock]
Meter=String
MeasureName=MeasureCoreClock
AutoScale=2k
Text=Core: %1Hz

; string — must be a String meter
[MeasureGpuName]
Measure=Plugin
Plugin=NativeHardwareMonitor
Category=GPU
Metric=Name
[MeterName]
Meter=String
MeasureName=MeasureGpuName
Text=%1

Units at a glance

  • Clocks: Hz (converted from MHz — use AutoScale=2k).
  • Sizes / throughput: bytes (or bytes/sec).
  • Percentages: 0–100.
  • Network link speed: bits/sec.

Full per-metric detail: Metrics Reference. For what -1 / -2 mean and common surprises: Quirks & Troubleshooting.

Clone this wiki locally