Skip to content

BENCHMARKS.md

Latest commit

 

History

History
109 lines (78 loc) · 2.86 KB

BENCHMARKS.md

File metadata and controls

109 lines (78 loc) · 2.86 KB

Benchmarks

This document records the benchmark scenarios, assumptions, and committed result snapshots used for Rezi. The numbers are directional, version-specific, and host-specific.

Purpose

The benchmark suite exists to answer practical engineering questions:

  • does a change regress Rezi on representative workloads?
  • how does Rezi behave on primitive, terminal, and structured-application scenarios?
  • how do changes affect throughput, latency, and memory on the same host?

It is not intended to serve as a universal leaderboard.

Active comparison set

The maintained suite currently compares Rezi against:

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

These systems differ in runtime, abstraction level, and language. Some are full UI frameworks, while others are lower-level terminal libraries. Absolute comparisons should be read with that context in mind.

Scenario groups

Primitive workloads

  • startup
  • tree-construction
  • rerender
  • content-update
  • layout-stress
  • scroll-stress
  • virtual-list
  • tables
  • memory-profile

Terminal-level workloads

  • terminal-rerender
  • terminal-frame-fill
  • terminal-screen-transition
  • terminal-fps-stream
  • terminal-input-latency
  • terminal-memory-soak
  • terminal-virtual-list
  • terminal-table

Full-application workloads

  • terminal-full-ui
  • terminal-full-ui-navigation
  • terminal-strict-ui
  • terminal-strict-ui-navigation

Interpretation

  • Rezi is designed for structured terminal applications where layout, routing, focus, and composition are part of the rendering cost.
  • Lower-level libraries may be faster on narrow output-only scenarios because they do less work per frame.
  • Memory and latency should always be read per scenario, not as a single global ranking.

Running the suite

Build prerequisites:

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

Quick all-framework run:

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

Rezi-only run:

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

Rigorous terminal run:

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

Artifact directories

Committed artifacts live under benchmarks/. They are retained as snapshots of specific runs and should be interpreted alongside the host, runtime, and suite configuration used to produce them.

Validity

See BENCHMARK_VALIDITY.md for the current assumptions behind the benchmark runner and result interpretation.