Benchmarks

Rezi includes a benchmark suite comparing terminal UI pipelines:

• Rezi (native): @rezi-ui/core via ui.* VNodes
• Ink: ink (React + Yoga + ANSI output)
• OpenTUI: @opentui/core + @opentui/react (Bun runner integration)
• Bubble Tea: github.com/charmbracelet/bubbletea (Go runner integration)
• blessed: imperative Node terminal UI baseline
• ratatui: native Rust baseline

The authoritative benchmark write-up and the latest committed results live in the repository:

• BENCHMARKS.md
• benchmarks/ (structured JSON + generated Markdown)
• Latest rigorous terminal-suite dataset: benchmarks/2026-02-19-terminal-v3/
• Latest OpenTUI React matchup dataset: benchmarks/2026-02-20-rezi-opentui-react-all-quick-v6/
• Latest OpenTUI Core matchup dataset: benchmarks/2026-02-20-rezi-opentui-core-all-quick-v4/
• Latest Bubble Tea matchup dataset: benchmarks/2026-02-20-rezi-opentui-bubbletea-core-all-quick-v3/
• GitHub: https://github.com/RtlZeroMemory/Rezi/blob/main/BENCHMARKS.md

Running benchmarks

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

node --expose-gc packages/bench/dist/run.js --output-dir benchmarks/local

For a faster smoke run:

node --expose-gc packages/bench/dist/run.js --quick --output-dir benchmarks/local-quick

Rezi vs OpenTUI across the full scenario set:

node --expose-gc packages/bench/dist/run.js \
  --matchup rezi-opentui \
  --io pty \
  --output-dir benchmarks/local-rezi-opentui

Rezi vs OpenTUI core-imperative (no React adapter):

node --expose-gc packages/bench/dist/run.js \
  --matchup rezi-opentui \
  --opentui-driver core \
  --io pty \
  --output-dir benchmarks/local-rezi-opentui-core

Rezi vs OpenTUI vs Bubble Tea (same scenario set, PTY mode):

REZI_GO_BIN=/path/to/go \
node --expose-gc packages/bench/dist/run.js \
  --matchup rezi-opentui-bubbletea \
  --opentui-driver core \
  --io pty \
  --output-dir benchmarks/local-rezi-opentui-bubbletea

Optional PTY mode (real TTY path; requires node-pty):

npm i -w @rezi-ui/bench -D node-pty
node --expose-gc packages/bench/dist/run.js --io pty --quick --output-dir benchmarks/local-pty-quick

Terminal competitor suite (PTY mode):

cd benchmarks/native/ratatui-bench
cargo build --release
cd -

node --expose-gc packages/bench/dist/run.js \
  --suite terminal \
  --io pty \
  --replicates 7 \
  --discard-first-replicate \
  --shuffle-framework-order \
  --shuffle-seed local-terminal-v3 \
  --cpu-affinity 0-7 \
  --env-check warn \
  --output-dir benchmarks/local-terminal-v3

OpenTUI rows are executed through packages/bench/opentui-bench/run.ts, which requires Bun. Default driver is react; use --opentui-driver core for imperative-core mode.

bun --version

Bubble Tea rows are executed through packages/bench/bubbletea-bench/main.go, which requires Go:

go version

Recommended: Go >= 1.24.

Interpreting results

See BENCHMARKS.md for:

• Terminal scenario definitions (terminal-rerender, frame-fill, screen-transition, fps-stream, input-latency, memory-soak, virtual-list, table)
• Methodology notes (process isolation, replicates, CI bands, shuffle order, CPU affinity)
• CPU/wall/memory interpretation and precision limits