What the Bottleneck Index measures
The Index measures balance, not speed. It summarises, for a set of common CPU-GPU pairings, which part usually limits performance at 1080p, 1440p and 4K, and how that balance shifts as new CPUs and GPUs launch. A pairing that is CPU-limited at 1080p and GPU-limited at 4K is not scored as good or bad in one number; its pattern across resolutions is the point, because a single number hides exactly the change that matters to builders.
That is why the Index is described as a quarterly summary rather than a calculator result. The calculator answers a personal question, which pairing limits your games, while the Index answers a market question, how balance is shifting across many pairings over time. Both use ranges and resolution patterns, and neither should be read as a single exact percentage or a guaranteed frame rate.
How the Index is built
The Index is built in three steps, each of which must be documented before numbers are published. First, a fixed set of pairings is chosen, covering common current and older CPUs and GPUs, so that each quarter compares like with like rather than changing the basket to suit a story. Second, each pairing is assessed by resolution and game type, using calculator data and public benchmark-scale data, with provisional values labelled as provisional. Third, changes are reported against the previous quarter, with the method and pairing set noted, so a reader can tell whether a change reflects new hardware or a changed method.
| Building step | What must be documented | Why it matters |
|---|---|---|
| Fixed pairing set | Which CPUs and GPUs, and why | Stops the basket changing to suit a story |
| Resolution and game type | 1080p, 1440p, 4K; esports, AAA, simulation | Stops one number hiding a resolution change |
| Change reporting | Previous quarter comparison, method notes | Shows whether hardware or method changed |
That discipline is the whole value of an index. Without a fixed pairing set and a documented method, a quarterly number is just a headline. With them, it becomes a record builders can trust, and a dataset other sites can check. Our method article explains the calculator model behind this approach, including its limitations.
How to read a quarterly update
A quarterly update will show three things: which pairings changed balance, and in which direction; which new CPUs and GPUs entered the set, and how they compare with the parts they replace; and which resolutions or game types drove the change. Read the direction and the reason together. A shift towards GPU limits at 1080p, for example, usually means CPUs have gained headroom or GPUs have gained speed, while a shift towards CPU limits usually means the opposite, and the update should say which.
Do not read a quarterly update as a buying list. An index movement for a whole pairing set does not mean your pairing changed, because your games, resolution, memory and cooling decide your own limit. Use the update to understand the market, then run your exact pairing in the PC bottleneck calculator and confirm it with your own monitoring test, as the 10-minute CPU bottleneck check explains.
What the Index is for, and what it is not for
- For builders: a sense of whether new launches are shifting balance towards CPU or GPU limits, and whether an upgrade class is becoming better value.
- For reviewers and sites: a citable, open dataset with a documented method, rather than a single percentage to quote out of context.
- Not for: judging one specific PC, which needs its own calculator result and monitoring test, or predicting a future launch, which no index can do.
The open dataset matters here. A quarterly index published with its pairing set and method, in a reusable format, can be checked, corrected and built on, while a closed number cannot. That openness is also what makes the Index a linkable asset: sites cite data they can verify, and ignore numbers they cannot.
What happens next
The next step is to publish the first measured update with its full pairing set, method notes and dataset, rather than to publish a placeholder number now. Until that update exists, this explainer is the correct record: it states the purpose, the method and the reading guide, and it deliberately contains no index values, because none have been measured yet.
Limitation: this article is an explainer, not a data release. It contains no Bottleneck Index numbers, quarterly values or measured results, because the first measured update has not been produced. Any future update should be read with its method notes, and any result should be treated as an estimate, not a guaranteed frame rate.
Frequently asked questions
What is the Bottleneck Index?
It is a planned quarterly summary of how CPU-GPU balance is changing across a fixed set of common pairings, by resolution and game type. It measures balance, not raw speed, and it is separate from the personal result the calculator gives for one pairing.
Why does this article contain no index numbers?
Because the first measured update has not been produced yet. Publishing invented numbers would defeat the purpose of an index, which is to record measured change over time with a documented method and a fixed pairing set. The first measured update, with its pairing set and method notes, is the correct next step.
How is the Index different from the calculator?
The calculator answers a personal question, which part limits your pairing in your games, while the Index answers a market question, how balance is shifting across many pairings over time. Both use ranges by resolution, and neither gives a single exact percentage. Both use ranges by resolution, and neither gives a single exact percentage.
How should I read a quarterly update?
Read the direction and the reason together: which pairings changed balance, which new parts entered the set, and which resolutions or game types drove the change. Do not read it as a buying list for your specific PC. Your own pairing should still be checked in the calculator at your own resolution.
Will the Index data be open?
That is the plan: a quarterly update published with its pairing set, method notes and a reusable dataset, so other sites and readers can check it. A closed number cannot be verified, and an unverifiable number is not useful as an index. That openness is what makes the Index citable, rather than just a headline number.