Labs

[!WARNING] Labs currently only supports Node.js. Workers are spawned via tsx with V8-specific flags (--allow-natives-syntax, --expose-gc), which are not portable to Bun (JSC) or Deno. Portability will be on the roadmap.

Labs is JS benchmarking you can trust. Trying to get good signal is harder than you might think. VMs are non-deterministic, environments are unstable, and typical benchmarks don't give you any sense of a comparison's validity. Labs detects variance, giving feedback on how to fix it, and uses statistical analysis to determine if two runs are actually different.

npm i @pmndrs/labs

Write

Create a config and a bench file. Benches use a generator where code before yield is setup, the yielded function is measured, and code after is teardown.

// labs.config.ts
import { defineConfig } from '@pmndrs/labs'

export default defineConfig({
  benchDir: '.',
})

Use @tags in the name string for filtering.

// array-push.bench.ts
import { bench, group } from '@pmndrs/labs'

group('array @stress', () => {
  bench('push 1k', function* () {
    const arr: number[] = []
    yield () => {
      for (let i = 0; i < 1000; i++) arr.push(i)
    }
  })
})

Run

Run your benchmarks with the bench command.

# Run all benches, save named after the current commit
bench
# Or filter by tag
bench "@mytag"
# Save with a name for easier reading
bench "@mytag" -n 'v1.0.0'

And get the pretty results.

labs

▶ relation-churn.bench.ts
clk: ~4.32 GHz
cpu: Apple M4 Pro
runtime: node 25.8.0 (arm64-darwin)

benchmark                   avg (min … max) p75 / p99    (min … top 1%)
------------------------------------------- -------------------------------
• relation churn
------------------------------------------- -------------------------------
■ big test                    17.80 ms/iter  18.02 ms      ▃▃██▃ ▃ ▃▆▃
                      (17.25 ms … 19.10 ms)  18.45 ms ▄███████████████▁▄▄▄▄
                  gc(  1.09 ms …   3.12 ms)   1.58 ms
                heap( 41.19 mb …  50.10 mb)  47.63 mb/iter

• avg/iter p75: Average time per iteration and p75, this is the most useful top metric.
• (min … max) p99: Fastest, slowest, and tail time that 99% of samples finish within. This shows the distribution visualized by the histogram.
• gc(min … max) p50: Per-sample garbage collection time above the fixed cost of collecting the process's live heap. Near zero means the bench adds nothing for the collector to do. Higher values mean it creates garbage or keeps more objects alive.
• heap(min … max) p50/iter: Bytes allocated per iteration, including typed array stores held outside the JS heap. The median ignores the allocation V8 does while compiling the function, so a bench that stays on integers and reuses its objects reports zero.

Compare

Compare against a baseline.

bench compare

And see the results!

━━ compare 2026-03-20_16-25-36 -> 2026-03-20_16-36-12
Apple M4 Pro
Mann-Whitney U on block medians  α=0.05  minΔ=5%

relation-churn.bench.ts
  bench                                    baseline  candidate    Δp50    Δp99     p       Δ 95% CI
---------------------------------------------------------------------------------------------------
  • relation churn
  -------------------------------------------------------------------------------------------------
  ▲ big test                                17.96ms    16.15ms  -10.1%   -9.8% <.001    -11.4..-8.8%
                                         ▁▂▄▅▅█▇▅▅▃ ▃▅▅██▅█▄▂▂

Guarantees

Labs promises to give results you can trust. To do this a number of guarantees are made when running benches.

• Each bench block runs in its own isolated worker process, preventing benches in the same file from contaminating each other's JIT state, heap layout, or GC history. Reordering benches can otherwise skew results by 2x or more.
• GC influence is mitigated. By default, each sample starts with a garbage collection (GC) reset so previous samples don't affect it.
• Every bench is run in isolated blocks that are interleaved: A₁ B₁ C₁ → A₂ B₂ C₂ → … → A₈ B₈ C₈. This reduces bias from gradual changes such as CPU throttling or even boosting.
• Timing overhead is controlled. If a sample's measurement time is so fast that the overhead of the timing itself would bias the results, then it is run in a batch sized so each sample lasts about a millisecond.
• Detect dead code elimination (DCE). If the samples measure the same as an empty function call, then we detect DCE and report it. This can be mitigated by returning a result from the yielded function.
• Detect noisy samples. Each block runs for a fixed time budget and sample floor. When the samples inside one process vary widely, a warning is given. This likely means the bench is non-deterministic or affected by runtime interference like background processes.
• Machine stability is measured. When median timings vary too much across blocks of samples, a warning is given. This can indicate an unstable machine ranging from thermal throttling to background processes.

Test correctness

A faster result only matters if the work is still correct. Labs keeps the result of the first untimed call and hands it back from yield once measurement finishes, so teardown can check it. Use any assertion library, or the small assert included with Labs. A failed check exits with code 1, keeps the saved result out of the baseline, and stops the remaining blocks of that bench.

import { assert, bench } from '@pmndrs/labs'

bench('sort', function* () {
  const result = yield () => [3, 1, 2].toSorted()
  assert.equal(result, [1, 2, 3])
})

Return the result from the generator to compare it automatically against the baseline's output.

bench('sort', function* () {
  const result = yield () => [3, 1, 2].toSorted()
  return result
})

For work that mutates state, use a snapshot hook to select the output.

bench('append', function* () {
  const values: number[] = []
  yield {
    bench: () => values.push(values.length),
    snapshot: () => values,
  }
})

When you compare runs, Labs replaces the speed verdict with output changed if the candidate no longer matches the baseline, and exits with code 1. Numeric outputs are compared with snapshotTolerance so float refactors do not trip it, non-finite numbers are spelled out and compared exactly, and anything else is digested. Assertions and snapshots run outside the timed work, so checking an answer does not affect its result.

How to control my CPU

One of the largest sources of noise when running benchmarks is an unstable environment, and the usual culprit is the CPU. The CPU boosts or thermal throttles, or a process gets put on a P-core (performance) instead of an E-core (efficiency). Labs checks the CPU clocks before and after each benchmark file and before each block, and tracks whether they vary across the runs. If it detects too much variance, you will get warned and the run will be flagged. But what can you do about it?

Windows

You get decent control with Windows by going into the BIOS. 90% of the variance is solved by disabling any kind of CPU turbo.

• Boot into the BIOS and disable turbo + SMT.
• Run the benchmarks on the highest priority.
• Disable as much background tasks as you can.

MacOS (Apple Silicon)

While Apple Silicon is relatively stable, there isn't much that can be done to control it. The governor cannot be adjusted and the dynamic CPU frequencies cannot be disabled.

Linux

Linux gives the most controls getting the best possible environment for testing. See this LLVM guide for specifics.

[!CAUTION] Below are AI generated docs that will get edited eventually. For now it lives here as notes.

API

Every run saves results by default, named after the current commit (abc1234, with -dirty when the tree has uncommitted changes, and a counter for repeat runs: abc1234-2). Outside a git repo, names fall back to a timestamp. Each result also records the commit, branch, and dirty state it was produced from. bench run is an alias for bench; pass --no-save to execute without saving.

pnpm bench                              # run all, save named after the current commit
pnpm bench "relation"                   # partial match on file name, save
pnpm bench "relation churn"             # separator-agnostic match, save
pnpm bench "@relation"                  # filter by tag, save
pnpm bench "churn @relation"            # name + tag combined, save
pnpm bench -n "v1.2.0"                 # save with explicit name (prompts if exists)
pnpm bench -n "v1.2.0" -f              # overwrite existing without prompting
pnpm bench -n "v1.2.0" --force         # same as -f
pnpm bench -n "v1.2.0" -m "refactor"   # save with name and description
pnpm bench --baseline                   # save and set as baseline
pnpm bench -b                           # shorthand for --baseline
pnpm bench -n "v1.2.0" -b              # save with name and set as baseline
pnpm bench --compare                    # save, then compare vs baseline
pnpm bench --blocks 12                  # save with 12 fresh-process blocks per benchmark
pnpm bench -c                           # shorthand for --compare
pnpm bench --last                       # rerun previous selection, save

Results are saved to <benchDir>/.labs/results/<name>.json and include hardware metadata (CPU, arch, runtime) for like-for-like comparisons.

Running without saving

pnpm bench --no-save                    # run all, no save
pnpm bench run --no-save "relation"    # filtered, no save
pnpm bench --no-save "@relation"       # filtered by tag, no save
pnpm bench --no-save --blocks 3        # no save, with 3 blocks for faster feedback
pnpm bench --no-save --last            # replay last selection, no save

Managing Results

pnpm bench list                        # list all saved results
pnpm bench delete "v1.2.0"             # delete a specific saved result
pnpm bench prune                       # remove results with unstable CPU clocks
pnpm bench clear                       # delete all saved results

bench list shows each result's name, description, timestamp, and CPU. The current baseline is marked with (baseline).

Baseline

pnpm bench baseline                    # interactive baseline picker
pnpm bench baseline "v1.2.0"          # set a result as the baseline
pnpm bench --baseline                 # save and set the new result as baseline
pnpm bench -b                         # shorthand for --baseline

Comparing

pnpm bench compare                     # interactive picker (latest preselected)
pnpm bench compare "v1.3.0"           # compare named result vs baseline
pnpm bench compare --last             # replay the last compared pair
pnpm bench compare -l                 # shorthand for --last

Outputs a colored table for each eligible benchmark:

| Column | Description | | --------- | ---------------------------------------------------------------------------------------------------------------------- | | baseline | Median of the baseline's fresh-process block medians | | candidate | Median of the candidate's fresh-process block medians | | Δp50 | Signed percent change between those two medians; descriptive and not used by the verdict gate | | Δp99 | Descriptive percent change in p99 from the pooled inner samples | | p | Two-sided Mann-Whitney U p-value on block medians; at or below alpha passes the statistical-significance gate | | Δ CI | Nominal 1 − alpha interval for the Hodges-Lehmann relative effect used by the verdict; not an interval around Δp50 |

Each row is prefixed with a verdict icon: green ▲ (faster), red ▼ (slower), or gray ■ (neutral). A red ✗ marks a bench whose snapshot differs from the baseline and shows output changed in place of a verdict. Candidate runs that failed a check or threw are listed under failed and receive no verdict. The verdict uses the Mann-Whitney p-value and the Hodges-Lehmann relative effect, not Δp50. Below each row, two distribution sparklines sit under their respective columns — baseline (cyan) and candidate (magenta) — on a shared axis. The sparklines use pooled inner samples and are descriptive only.

Comparison is gated. Two runs must pass environment checks before results are shown.

Environment checks (fail = entire comparison is denied):

• Hardware match — CPU model, architecture, and runtime (Node/Bun/etc.) must be identical between runs.

Environment warnings (non-blocking, printed above results):

• Clock drift — if either run's CPU frequency drifted > 5% during the run, a warning is shown. On Apple Silicon this is expected (no governor or turbo control); on other platforms it usually means turbo boost or thermal throttling is active.
• Clock speed mismatch — if the two runs' median clock speeds differ by > 5%, a warning is shown. Absolute timings may not be directly comparable.
• Block count mismatch — unequal counts are supported, but the actual counts determine whether the test can reach the configured significance level.

Per-benchmark eligibility and annotations:

• Not missing — the bench must exist in both runs. Benches present only in baseline or only in candidate are reported separately.
• Limited-resolution benches are annotated, not skipped — a bench whose approximate between-block resolution is coarser than minDelta still gets judged because the rank test already responds to spread. Its row carries a ⚠ ~±N% marker; a neutral result there is inconclusive at the shown resolution, not evidence of no change.
• Block replication — both sides need at least two fresh-process blocks, and their combined counts must permit an exact p-value at or below alpha. Results saved before blocked sampling are skipped.

Writing a bench

import { bench, group } from 'labs'

group('my-group @mytag', () => {
  bench('my-bench', function* () {
    // setup
    yield () => {
      const result = /* measured computation */ 1 + 1
      return result
    }
    // teardown
  })
})

Dead Code Elimination

Engines may remove pure computations whose results are never used, making a benchmark appear impossibly fast. When benchmarking pure work, return its result; Labs consumes non-undefined returns from automatically timed function benchmarks inside the timed region. The returned value must depend on the measured work—no return is needed when the work already has observable side effects.

Tags

Tags are @-prefixed tokens in the group or bench name string. They are stripped from the display name and used for filtering.

group('relation-queries @relation', () => {
  bench('ChildOf(parent)', function* () { ... });
  bench('wildcard @slow', function* () { ... }).gc(false);
});

Tags inherit: ChildOf(parent) has effective tags [@relation], wildcard has [@relation, @slow].

Filter by tag (quote the @ so pnpm passes it through):

pnpm bench "@relation"    # runs both benches
pnpm bench "@slow"        # runs only wildcard

Statistical comparison strategy

Every run measures each benchmark in fresh-process blocks, eight by default. The first block of each benchmark runs until both the block time budget and the sample floor are met, then records its batching and sample-count decisions; later blocks replay that plan exactly. Blocks are interleaved across benchmarks so every benchmark spans the run. Inner timing samples remain useful for distributions and p99, but each block's median is treated as one independent experimental unit for comparison verdicts.

A change is flagged only when both conditions are met:

1. p ≤ alpha (two-sided Mann-Whitney U, default 0.05) — statistical significance across block medians. Comparisons with at most 50 blocks combined use the exact conditional permutation distribution of the observed ranks, including tied medians. Larger samples use a continuity- and tie-corrected normal approximation.
2. |Hodges-Lehmann Δ| ≥ minDelta (default 0.05) — practical magnitude. The relative estimator is the median of all pairwise candidate / baseline block-median ratios, minus one. Positive values mean slower and negative values mean faster.

Both gates must pass to report faster or slower; otherwise the result is neutral. The displayed Δp50 is the percent change between the two median block medians, so it can differ slightly from the Hodges-Lehmann effect used by the verdict. The Δ CI column is a rank-based interval around the Hodges-Lehmann relative effect. Its endpoints use exact tie-free Mann-Whitney critical ranks at the nominal 1 − alpha level; ties make the interval slightly conservative. The interval is uncertainty context, not a third verdict gate and not a confidence interval for Δp50.

Effect-size gating (Cliff's d) was removed: on block medians it is a monotone transform of the same U statistic behind the p-value, so a separate threshold added confusion without adding information.

A verdict must also survive the clock cross-check: when the two runs' median block clocks differ by more than 2%, the same block medians are re-judged in estimated CPU cycles (median × its block's clock probe), and a disagreement between the time and cycles verdicts skips the bench as clock-confounded rather than reporting a shift that may just be a frequency difference. When clocks are effectively equal the cross-check is inert — cycles would only re-scale time by probe jitter.

The pooled inner-sample sparklines and p99 ratio are descriptive. They help expose distribution and tail changes but are not independently significance-tested.

Between-block spread and resolution

Labs summarizes each run's block medians with a robust relative spread: 1.4826 × MAD / median. It turns that spread into an approximate minimum detectable effect using 2.8 × spread × √(2 / blocks). The estimate is a normal-theory planning heuristic for 5% significance and 80% power with equal-sized groups, while actual verdicts use the rank test above. Treat it as an order-of-magnitude resolution diagnostic, not a guaranteed or hard detection limit.

During comparison, Labs calculates the resolution separately for baseline and candidate and displays the worse of the two. When it exceeds minDelta, the row is annotated as limited-resolution, but the verdict is still evaluated. Large, well-separated effects can therefore receive a verdict despite inconsistent fresh runs; a neutral result at limited resolution means that the data could not establish a change at that scale.

Each p-value applies to one benchmark. Labs does not currently adjust alpha across a suite, and separately saved baseline and candidate sessions can still differ in unmeasured machine state. Treat suite-wide or causal conclusions accordingly.

Config

Place labs.config.ts alongside your bench files:

import { defineConfig } from 'labs'

export default defineConfig({
  benchDir: '.',
  benchMatch: '**/*.bench.ts',
  nodeFlags: ['--allow-natives-syntax', '--expose-gc'],
})

| Option | Default | Description | | ------------------- | ------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------- | | benchDir | (required) | Directory to search, relative to config file | | benchMatch | **/*.bench.ts | Glob pattern for discovery | | nodeFlags | ['--allow-natives-syntax', '--expose-gc'] | Node flags per worker process | | resultsDir | .labs | Directory for saved results, relative to config | | blockTime | 0.5 | Time budget per block in seconds; every block runs at least this long and collects at least minSamples | | minSamples | 20 | Minimum samples per block | | alpha | 0.05 | Mann-Whitney U significance level | | minDelta | 0.05 | Minimum absolute Hodges-Lehmann relative effect for a verdict; rows whose approximate resolution exceeds it are annotated as limited-resolution | | snapshotTolerance | 1e-9 | Relative tolerance for numeric snapshots, with an absolute floor at one; digests of non-numeric snapshots are compared exactly | | blocks | 8 | Fresh-process blocks per benchmark; override per run with --blocks |

Sampling behavior:

• The first block of each benchmark samples until both blockTime and minSamples are met. Fast work is batched so each sample lasts about a millisecond, and slow work runs until the sample floor is reached even if that exceeds blockTime.
• Later blocks replay the first block's batching and sample count exactly, so every block does identical work.
• Two spreads are reported. Between-block spread of medians drives the comparison resolution and the limited-resolution annotation, evaluated against the current minDelta so changing it re-evaluates existing results. Within-block spread of samples flags noisy benchmarks whose timed work does not settle inside one process.