Scenario & Run Center
Define target, build type, warmup and sample policy, repetitions, cooldown, transport, environment requirements, metrics, markers, loading transitions, baselines, and budgets.
Performance Lab
Coming soonCompare repeatable performance scenarios, statistical baselines, regression budgets, and build reports.
Marketplace release coming soon
In development · 26.9.0
one connected workflow
The product is designed around evidence validity rather than a single FPS number. Missing samples, unsupported capabilities, environment mismatches, interrupted captures, and invalid evidence remain explicit non-passing states.
Define target, build type, warmup and sample policy, repetitions, cooldown, transport, environment requirements, metrics, markers, loading transitions, baselines, and budgets.
Build once for desktop runs, bind the exact player inventory by SHA-256, and verify those bytes before every repetition.
Review minimum, median, p90, p95, p99, maximum, raw/included/excluded counts, and an explicit none or reviewed Tukey outlier policy.
Use baseline/candidate, strict A/B, or contiguous block-balanced A-B-B-A comparisons with immutable source identities and environment checks.
Inspect completed build result, duration, size, file inventory, warnings, errors, options, environment, and reviewed budgets from the built-in Performance Lab window.
Export JSON, Markdown, HTML, JUnit, SARIF, Release Core evidence, and portable external-device bundles through the same production services used by the Editor UI.
development status
Local desktop orchestration supports macOS, Windows, and Linux code paths. Android, iOS/iPadOS, and WebGL use immutable external capture kits. Exact 26.9.0 source validation passes on macOS, Windows, and Linux, but frozen-candidate real-player and physical-device matrices remain open, so this page does not promote those workflows to release-validated status.
Version 26.9.0 artifact-readiness builds were byte-identical and their Asset Store containers clean-imported in both supported parent modes. Release eligibility still requires candidate freeze, the full player, transport, lifecycle, coexistence, and paired-product matrices, final marketplace audits, and official Unity Recorder storefront media.
local and reviewable
Performance measurements, baseline history, BuildReport evidence, and exported reports stay local to the project workflow. The package does not require a cloud account or background performance-data upload.