Quickstart

The numbers on this page come from a real run: the engine module of opengrid, loaded and queried with 100 000 rows.

1. Describe the module

Put peakmark.toml next to the baseline it will keep. Paths in it are relative to the file.

[[module]]
name     = "engine"
artifact = "pkg/engine_bg.wasm"
glue     = "pkg/engine.js"

[module.budget]
brotli         = "170 kB"
peak_memory    = "26 MiB"
growth_per_run = "0"

[[scenario]]
module = "engine"
file   = "bench/pivot.mjs"
repeat = 5

2. Write a scenario

A scenario is an ES module whose default export gets the loaded module and drives it the way a page would. See Writing a scenario.

3. Accept the first measurement

Until something is accepted, peakmark check warns that there is no baseline and names the command:

peakmark accept --all --reason "first measurement"

4. Check on every change

peakmark check

Within budget and baseline it passes. Over the budget it exits with 1 and names the scenario that reached the value; here with peak_memory lowered to 20 MiB:

✖ peakmark check: over budget: engine

  Modules: 1   Over budget: 1   Grown: 0

● engine (1)
  - Peak memory 23.6 MiB, budget 20.0 MiB [mem/peak] at bench/pivot.mjs

Next steps:
  • Shrink the module, or raise its budget in the configuration on purpose

Raise the budget in peakmark.toml on purpose, or shrink the module. Growth that stays under the budget but passes the baseline is only a warning; see Budget and baseline.

In a pipeline

peakmark check exits with 0 within budget, 1 over budget and 2 if it could not run, so it works as a step in any CI. --json prints the result for other tools.