Methodology and sources

Every FPS figure on MacGameFPS comes from a formula with five public ingredients. This page explains them and shows how far the formula lands from real measurements.

1. The graphics power of each chip

We use the Geekbench 6 Metal score of every Apple Silicon chip (21 chips, from M1 to M5 Ultra). It is the same test for all of them, so it compares across generations. Geekbench averages every GPU configuration of a chip, so we scale the score to each model’s exact GPU core count.

Provisional: M5 Ultra and M6 just launched and are not yet on Geekbench’s chart. We use their first published results and will update them once an average exists.

2. One anchor point per game

Each game has a reference point: the FPS it gets on a specific chip, resolution and preset. When a reliable published measurement exists we use it; otherwise we start from the studio’s official requirements (for example, “M1 with 16 GB for 1080p at 30 FPS on Low”). The formula scales from there to every other Mac.

3. Resolution and quality

FPS drop a little less than pixel count when resolution rises (some per-frame work does not depend on it), and each preset has an average cost: Medium runs at about 86% of Low, High 71% and Ultra 59%.

4. What is specific to the Mac

• Cooling: a MacBook Air has no fan and loses up to 20% of sustained performance in demanding games. MacBook Pros lose 5% versus a desktop with the same chip.
• Unified memory: if the Mac has less memory than the game’s minimum, we give no figure — the game will not launch or is not playable. Right at the minimum, the heaviest games lose a little to memory pressure.
• MetalFX: Apple’s upscaler renders at a lower resolution. We estimate +24% FPS in Quality mode, +50% in Balanced and +70% in Performance, when the game offers it.

5. The engine ceiling

No game scales forever with the GPU: at some point the CPU or the engine itself takes over. Each game has an FPS ceiling and the formula approaches it smoothly.

Validation against real measurements

We compare the formula with 136 measurements published by specialist outlets. The mean absolute error is 17%. Figures labelled “measured” on the site are those measurements, linked to their source.

Honest limitations

• These are estimates: the mean error against real measurements is 17%, and individual cases can exceed 25% (very busy scenes, CPU-bound games or recent patches that change performance).
• Many games have no published Mac measurements: for them we start from official requirements and similar games. They carry the “Estimate” label.
• We do not model ray tracing or frame generation: every figure is without ray tracing.
• Real performance varies with game and macOS versions, room temperature and whatever else you have open.

How often we update

We review the data when a new Mac or chip ships, when a studio changes its requirements and when reliable new measurements appear. Last review: October 1, 2026.

Sources

• Geekbench Browser — Metal Benchmarks
• Apple tech specs for each model’s memory, GPU cores and cooling.
• Official Mac requirements from each studio or porting house (linked on every game).
• 228 measurements published by specialist outlets (table above).

Photo credits

Mac product photos are by Apple Inc., from its press room, Apple Newsroom. They are used for information purposes, to identify each model:

• Introducing the next generation of Mac
• Apple introduces the new MacBook Air with the M4 chip and a sky blue color
• 13-inch MacBook Pro with M2 available to order starting Friday, June 17
• Apple unveils game-changing MacBook Pro
• Apple's new Mac mini is more mighty, more mini, and built for Apple Intelligence
• iMac features all-new design in vibrant colors, M1 chip, and 4.5K Retina display
• Apple unveils new Mac Studio, the most powerful Mac ever
• Apple unveils powerful, all-new Mac Pro and groundbreaking Pro Display XDR
• Say hello to MacBook Neo