By Walter Jobs, Technology writer and editor · Published 9 October 2026
Short answer: on its own, no. A percentage tells you the gap between two parts and nothing about whether that gap costs you anything you can see. A 25% figure that still leaves you above your monitor’s refresh rate is irrelevant. A 9% figure that drops you from 62 to 56 frames per second in the game you actually play is worth acting on. The number is a starting point, and most places that publish one stop there.
We publish a percentage ourselves, so this is partly an argument about how to read our own output.
What the number leaves out
Four things, each of which can flip the conclusion entirely.
The frame rate it happens at
A 20% gap at 180 frames per second means you could have had 216. On a 144 Hz display you would see neither. The same 20% at 50 frames per second means you could have had 60, which is the difference between uncomfortable and fine.
Identical percentage, opposite advice. This single omission accounts for most of the bad decisions people make from these tools.
Your monitor
Frames above your refresh rate are not shown. If your screen is 60 Hz and the calculator says you are leaving 30 frames per second on the table at 150, you are leaving nothing you could perceive. Any serious reading of a bottleneck figure has to start with what your display can actually present.
The resolution and the game
A pair that is 30% processor limited at 1080p can be perfectly balanced at 1440p, because the card’s workload quadruples while the processor’s barely moves. A figure quoted without a resolution attached is averaging over situations that disagree with each other. Genre does the same thing: the same pair behaves differently in a strategy game and a corridor shooter.
What fixing it would cost
A 15% processor limit that needs a new chip, board and memory is a different proposition from a 15% limit resolved by adding a second memory stick. The percentage is identical and the decisions are not remotely similar.
The other problem: everyone computes it differently
There is no standard definition. Run the same pair of parts through several calculators and you will get several different numbers, because each is measuring something slightly different and none of them say which.
| A percentage can mean | Which produces |
|---|---|
| The gap between two performance index scores | A large number that may cost no frames at all |
| Frames lost against a theoretical perfect partner | A number that depends entirely on which partner was chosen |
| How far one part sits outside a healthy ratio band | A smaller, more conservative number |
| Idle time on the faster component | Something closer to a utilisation figure than a performance one |
Those are four different quantities wearing the same label. A reader comparing a 40% figure from one site with a 12% figure from another is not comparing like with like, and neither site usually explains which it computed. Ours is the third of those, and the band thresholds and the formula are written out on how we calculate this, so you can see what the number is before you trust it.
The four questions that replace the percentage
Ask these instead. They can be answered from the same information and they lead somewhere.
| Question | Why it decides something |
|---|---|
| 1. What frame rate do I get now? | If it already exceeds your monitor, stop here |
| 2. What would I get with the limit removed? | The gap in frames is the thing you are buying |
| 3. Would I see the difference? | Compare against your refresh rate, not against zero |
| 4. What does removing it cost? | Frames per pound is the only honest comparison |
A worked example. Say a calculator reports 22% and you play at 1440p on a 144 Hz screen. You currently get 96 frames per second and the figure suggests 117 with a better processor. You would see that difference, since both are below 144. The upgrade needs a chip, a board and memory. So the real question is whether 21 frames is worth that bill, and that is a question you can actually answer. The 22% on its own was not.
When the percentage is genuinely useful
It is not useless, and it would be unfair to pretend otherwise. Three jobs it does well.
- Comparing candidates. Working out which of three cards suits the processor you already own is exactly what a relative figure is for.
- Finding the direction. Knowing which of your two parts is the limit tells you which half of the market to shop in, and that is most of the value.
- Spotting the extremes. A very large figure is a reliable signal that something is badly mismatched, even before you look at frame rates.
What it cannot do is tell you to spend money, because that decision needs a frame rate, a monitor and a price, and the percentage contains none of those. This is why our own result shows the figure alongside the frame rate at each resolution and names the specific upgrade, rather than presenting the percentage by itself. Run your parts through the bottleneck calculator and read the frame rate column first.
Frequently asked questions
What is a good bottleneck percentage?
There is no threshold that works as a rule, which is the honest answer. A figure is acceptable when the frame rate you get is enough for your display and your expectations. People quote 10% as a target, and it is arbitrary: 10% at 40 frames per second matters and 10% at 200 does not.
Why do different bottleneck calculators give different answers?
Because there is no agreed definition of what the percentage measures. One may report the gap between index scores, another the frames lost against a hypothetical ideal partner, another the distance outside a balanced ratio. All four are defensible and they produce different numbers from the same inputs. Check what a tool says it is computing before comparing its output with another’s.
Should I aim for a 0% bottleneck?
No, and it is not achievable anyway. The two halves of the pipeline respond to different variables, so a pair balanced at one resolution in one game will be unbalanced in another. Aiming for zero means overspending on whichever part is currently behind, and the balance will move the next time you change a setting.
Is a 30% bottleneck bad?
It depends entirely on where it lands. If it takes you from 190 to 145 frames per second on a 144 Hz display, you will never see it. If it takes you from 85 to 60, you will. Find the two frame rates before judging the percentage between them.
Are bottleneck calculators accurate?
They are models, including ours, and they should be read as estimates with a direction rather than as measurements. They cannot know your specific game, your settings, your memory configuration or your background load. What they do reliably is tell you which part is the limit and roughly how far apart the two are, which is enough to decide where to look next.
The figures used as examples here are illustrative rather than measured, chosen to show how the same percentage leads to opposite conclusions. Our own method, including what our percentage measures and where it stops being reliable, is on how we calculate this.