Benchmarks

Rezi includes a benchmark suite for checking render-path behavior across a fixed set of scenarios. The results are host-specific and directional. Use them as engineering data, not as a universal ranking.

The longer reference write-up and committed artifacts live in:

• BENCHMARKS.md
• benchmarks/

What the suite covers

• primitive workloads such as startup, rerender, content updates, layout stress, and virtualized lists
• terminal-level workloads that include PTY output
• full-application workloads with structured layouts and navigation

Active comparison set

The maintained suite compares Rezi against a small set of other terminal UI runtimes and lower-level libraries:

• OpenTUI (React)
• OpenTUI (Core)
• Bubble Tea
• terminal-kit
• blessed
• Ratatui

These are not equivalent systems. Some are higher-level UI frameworks, some are lower-level terminal libraries, and some run on different runtimes or languages. Read the numbers with that context in mind.

Current reading of the results

• Rezi is designed to stay competitive on structured terminal UI workloads where layout, routing, focus, and composition are part of the cost.
• Lower-level libraries can be faster on narrow buffer-write scenarios because they do less work per frame.
• Memory and throughput vary significantly by scenario, host, and runtime.

Running the suite

Prerequisites:

npm ci
npm run build
npm run build:native
npx tsc -b packages/bench

Quick run:

node --expose-gc packages/bench/dist/run.js \
  --suite all --io pty --quick \
  --output-dir benchmarks/local-all

Rezi-only quick run:

node --expose-gc packages/bench/dist/run.js \
  --framework rezi-native --io pty --quick

Interpretation rules

• compare runs only when the scenario, framework set, and host conditions match
• prefer repeated runs for decision-making
• treat committed benchmark files as snapshots, not promises

See BENCHMARKS.md for the fuller methodology note and artifact references.