Benchmarks

TL;DR: CEM’s Go implementation is significantly faster than JavaScript-based tools. Generate benchmarks show ~10x speedup over web-component-analyzer. LSP benchmarks demonstrate sub-millisecond response times for autocomplete and validation.

Overview

CEM provides two types of benchmarks for comprehensive performance analysis:

Generate Tool Benchmarks

Performance comparison of custom element manifest generation tools.

Number of runs per tool: 100
Number of files analyzed per run: 45

Generate tool benchmark results: lower is better (seconds)

@lit-labs/cli

Docs

1.47s

Avg Time

52.3KB

Avg Output Size

100

Runs

Valid

Validation

Command

npx --yes @lit-labs/cli labs gen --manifest --out data/lit

Run Breakdown: Time per Run (seconds)

Validation Results

No validation issues found - The generated manifest is valid and follows all best practices.

@custom-elements-manifest/analyzer

Docs

0.83s

Avg Time

66.1KB

Avg Output Size

100

Runs

Valid

Validation

Command

npx --yes @custom-elements-manifest/analyzer analyze --outdir data/cea --globs benchmark/components/*.ts

Run Breakdown: Time per Run (seconds)

Validation Results

No validation issues found - The generated manifest is valid and follows all best practices.

cem generate

Docs

0.34s

Avg Time

73.6KB

Avg Output Size

100

Runs

Valid

Validation

Command

cem generate -o data/cem/custom-elements.json benchmark/components/*.ts

Run Breakdown: Time per Run (seconds)

Validation Results

No validation issues found - The generated manifest is valid and follows all best practices.

LSP Server Benchmarks

Last benchmark run: 2026-05-10 18:59:35

HTML File Operations

Pure LSP protocol timing in .html files. Each benchmark measures the time from LSP request to response completion, using multiple iterations to calculate statistical distributions (mean, median, P95, P99).