Methodology

How Game Rig Lab sources and presents data — and what is and is not live yet.

Hardware capability model (v1)

Every ranked CPU and GPU on Game Rig Lab carries a relative capability index, derived from PassMark CPU Markand PassMark G3D Mark — broad-coverage, continuously-updated aggregate benchmark databases. Within each hardware type, scores are rebased so the weakest item in our own catalog equals 100; a product's index is never compared across CPU and GPU, and is never shown to you as a raw number — only as a restrained relative tier (Entry-level / Mainstream / High-end / Enthusiast).

This index answers relative-capability and compatibility-screening questions ("is this roughly at, above, or below what a game asks for?") — it is not an FPS predictor. PassMark's PerformanceTest suite is a general-purpose composite benchmark, not a game-specific one, so results are shown as coarse states (exceeds / meets / borderline / below / unknown) rather than a precise score. Hardware with no confidently-sourced benchmark data is left unranked — shown as "unknown," never guessed as a pass or fail. Separately marketed memory/SKU variants (e.g. different VRAM capacities) are always kept as distinct products with their own independently-sourced figures, never merged for convenience.

Full sourcing, normalization, and reconciliation rules live in this repository'sdocs/hardware-capability-methodology.md, versioned as v1.

Compatibility screening (Can I Run It / Upgrade Advisor)

Can I Run It and Upgrade Advisor are both thin applications of the same hardware capability model above — neither introduces a second score, an FPS estimate, or a price/value ranking. A component's status against a game's official minimum or recommended requirement is always one of five states: exceeds, meets, borderline, below, or unknown. Identical canonical hardware always meets itself. Official requirement text (including games with several co-equal official CPU/GPU alternatives, e.g. "RTX 3060 or RX 6600") is always shown exactly as published — matching ANY one satisfied alternative is sufficient, and the wording itself is never edited or replaced with a single chosen product.

Limiting components. A component marked "below" is definitely limiting; one marked "borderline" is only a possible limitation and is always described with hedged wording — never as "this can't run the game." Multiple components may be limiting at once (a strong GPU never excuses a CPU that's below requirement, or vice versa — this is component gating, not an averaged score). An "unknown" component is never resolved into a limiting/non-limiting conclusion; it simply stays unknown and never blocks a confident conclusion about a different, ranked component.

Upgrade candidates. When a component is below or borderline, Upgrade Advisor looks for real, non-fixture canonical products of the same type that (1) actually satisfy the requirement (meets or exceeds — a borderline candidate is never suggested as the fix) and (2) are not weaker than your current component. Results are ordered by smallest performance headroom above the requirement first — the smallest defensible upgrade step, never automatically the single strongest product in the catalog. Candidates are labeled "performance-suitable" deliberately: socket, motherboard, RAM generation, cooler, power supply, and case fit are not evaluated yet, so these are never presented as drop-in or guaranteed-compatible upgrades.

None of these tools make an FPS prediction, a "guaranteed playable/smooth" claim, or a price/value judgment — none of that evidence exists in this catalog yet.

PC Build Recommender

PC Build Recommender answers a different question than Upgrade Advisor: not "what should I change about my existing rig?" but "starting from nothing, what performance-suitable CPU/GPU/RAM core would clear this game's requirement?" It reuses the exact same candidate search as Upgrade Advisor — real, non-fixture, ranked candidates that actually satisfy the official requirement (meets or exceeds) — simply without a "current component" to compare against.

Lowest sufficient picks the smallest capability step in that ranked list — the least amount of hardware that defensibly clears the requirement. Balanced headroom picks a modestly stronger option further into the same ranked list — a fixed, documented position, not a fabricated "best" score — and is never the single strongest product in the catalog merely because it's the strongest. Multiple official requirement alternatives (e.g. "RTX 3060 or RX 6600") are respected exactly as Upgrade Advisor respects them: a candidate satisfying any one alternative qualifies.

The result is a performance-suitable core recommendation, explicitly not a complete, physically compatible build — this project has no socket, motherboard, RAM- generation, cooler, power-supply, or case-fit data yet, so wording like "fully compatible," "ready to assemble," or "drop-in" is never used. RAM is shown as the game's published target only; no DDR generation, speed, timing, brand, or kit is ever invented. Like every other tool here, no price or value ranking is applied — nothing in this catalog is evidence for that yet.