Plot resource-usage logs recorded by the autometric package

Render a ggplot of the CPU and memory usage recorded by autometric::log_start() / autometric::log_stop(). Each selected metric is drawn as a line over time, faceted so that percentage metrics (CPU) and megabyte metrics (memory) keep their own y-axis scale. This turns an autometric log into a publication-ready resource-usage figure and pairs naturally with the carbon-footprint estimates of ga_footprint().

Usage

ga_autometric_plot(
  log,
  metrics = c("cpu", "resident"),
  color_by = "phase",
  ...
)

Arguments

log

Either a path to an autometric log file (read with autometric::log_read()) or a data frame already returned by autometric::log_read() (with at least the columns time, cpu, core, resident, virtual).

metrics

(character, default c("cpu", "resident")) Which columns to plot. Any subset of "cpu" (\ "resident" (resident memory, MB) and "virtual" (virtual memory, MB).

color_by

(character, default "phase") Column used to color the lines, typically "phase", "name" or "pid" so several logged phases or processes can be compared on one plot.

...

Additional arguments passed on to autometric::log_read() when log is a file path.

Value

A ggplot object with one facet per metric.

See also

ga_targets_network(), ga_footprint()

Author

Adrien Taudière

Examples

if (requireNamespace("autometric", quietly = TRUE)) {
  log_file <- tempfile()
  autometric::log_start(path = log_file, seconds = 0.05)
  x <- numeric(0)
  for (i in 1:5) {
    x <- c(x, rnorm(2e5))
    Sys.sleep(0.1)
  }
  autometric::log_stop()
  ga_autometric_plot(log_file)
}