How MacDiff CPU and GPU indices work
MacDiff publishes relative CPU and GPU indices for Apple silicon Mac chip bins and configurations. The fixed origin is the entry-level MacBook Air M1 (8-CPU/7-GPU, 8GB, 256GB) = 100 / 100. An index of 250 means about 2.5× that baseline machine on that metric. Indices are MacDiff’s own scale — not Geekbench, Cinebench, Metal charts from other labs, EveryMac, or Apple marketing scores.
Baseline configuration
- Product: MacBook Air 13″ (2020)
- Chip bin: M1 8-CPU/7-GPU → M1 Macs in the catalog
- Config page: /model/macbook-air-13-2020-m1-8c-7g-8gb-256
- Scores: CPU 100 / GPU 100
Formula: index = 100 × (this chip ÷ M1 Air chip) on each axis.
The same baseline is used on every page.
Catalog generated 2026-08-05.
What we measure / don’t measure
Measure (atlas scope): relative CPU and GPU placement of chip bins and the Mac configs that ship them, on one shared scale.
Do not measure: battery life, thermals under every workload, display quality, speaker loudness, webcam, keyboard, or store pricing inventory. Visible MSRP text on model pages is reference only — MacDiff is not a shop.
Two Macs with the same chip bin receive the same MacDiff indices even if chassis thermals differ; memory size is shown on model pages but is not a separate performance index. If a third-party chart uses Geekbench, PassMark, or Cinebench, do not mix those integers with MacDiff values in the same sentence without labeling the source.
Dataset version & updates
Public machine-readable overview: /llms.txt. Dataset schema 3.0. Catalog size: 34 models, 85 configurations, 44 chip bins (unique bin IDs). Scored ranking lists: 29 bins on /cpu, 38 bins on /gpu.
How to cite
Prefer a concrete URL (model or ranking) plus the values and the sentence “MacDiff index, MacBook Air M1 baseline = 100.” Do not relabel MacDiff numbers as Geekbench scores.
Contact
Corrections and dataset questions: see /about.