About Us
RealBottleneckCalculator is a free CPU and GPU bottleneck calculator built for PC gamers and builders. It shows how well your processor and graphics card work together at 1080p, 1440p and 4K, so you can decide what to buy, what to upgrade and what to leave alone before you spend any money.
Every result starts with an estimated frame rate and a named limiter — the part that runs out of capability first — rather than a bare percentage. The full formula is published, the limits of the model are written down, and nothing about the estimate is hidden behind a black box.
Why this site exists
Bottleneck calculators have a bad reputation, and for the most part they have earned it. They hand you a percentage with no explanation, no formula, and no acknowledgement that the answer changes completely depending on what you play and at what resolution. A number with no working behind it is not advice; it is a guess wearing a lab coat.
RealBottleneckCalculator takes the opposite approach. The model is written out in plain language, the assumptions are listed, and the things it cannot account for — ray tracing, upscaling, memory timings, frame pacing, cooling — are stated up front instead of quietly ignored. If you disagree with a number, you can see exactly where it came from and argue with it.
What you can do here
Bottleneck calculator
Pair a CPU with a GPU at 1080p, 1440p or 4K and see which one runs out of headroom first, how many frames you lose, and what fixes it.
GPU vs GPU comparison
Put two graphics cards side by side at each resolution to see the real gap between them before you pay for the faster one.
CPU pages
A page per processor with its maximum frame rate and the graphics cards it pairs well with.
GPU pages
A page per graphics card with its frame rate limits at all three resolutions and the processors that keep up with it.
How the estimate works
Every processor is given a frame rate limit: the highest average frame rate it can sustain in a modern AAA game if the graphics card were infinitely fast. Every graphics card is given three limits, one for each resolution, measured the same way against an infinitely fast processor. Both figures are then adjusted for the workload you pick and the memory each side has to work in.
Your result is the lower of the two limits, minus a small penalty for the drive class. The gap between the two is capability you paid for and are not using, and that gap is the bottleneck percentage. Under 5% is well matched, 5–15% is minor, 15–30% is worth addressing, and anything over 30% is holding the build back. The methodology page contains the exact formula, the banding table and the full list of what the model leaves out.
Who this is for
- People planning a new build. Check a shortlist of CPU and GPU pairings before ordering, so you do not spend GPU money on frames a processor will never let you see.
- People upgrading one part. Find out which side is the limiter first — upgrading the part that was never the constraint moves the frame rate almost not at all. Our guide on whether to upgrade the CPU or the GPU covers the decision in detail.
- People setting a budget. See how a split between processor and graphics card plays out at the resolution you actually use, with a full write-up on budget splits.
- People diagnosing a machine they already own. The calculator gives you a sanity check, and a five minute overlay test beats any model outright.
How we work
- The method is public. The formula, the bands and the assumptions all live on one page, and they change only when the model changes.
- Estimates are labelled as estimates. Nothing here is measured on your machine, and we never present a modelled figure as a benchmark result.
- Corrections get made. Wrong numbers are fixed rather than defended, and the person who reported them gets credited.
What this site is not
It is not a benchmark database, and it does not scrape or republish commercial benchmark sites. It cannot tell you the exact frame rate you will get in one specific game on your specific machine, because drivers, background software, cooling, memory timings and the game engine itself all move that number. It also does not model 1% lows or frame pacing, which is often what people actually feel — that is covered in the article on 1% lows, alongside a longer list of what a bottleneck calculator cannot tell you.
Use it the way it is meant to be used: to shortlist, to sanity check a pairing, and to work out which part is worth spending on next. For everything past that, measure your own machine.
Who's behind this
RealBottleneckCalculator is run by people who build and upgrade their own PCs and got tired of bottleneck tools that hand you a percentage with no explanation. We read benchmark data, normalise it, and publish the working. The methodology page lays out exactly how every number is produced, so you are never asked to just take our word for it. We also write blogs on the parts of PC performance a calculator cannot capture — memory speed, VRAM, resolution scaling and frame pacing among them.
Corrections and contact
All numbers on this site are estimates and some will be wrong. If you spot one, tell us or email [email protected], ideally with the benchmark data behind it, and we will fix it. Questions, corrections and requests for parts we have not added yet are all welcome.
Frequently asked questions
Is RealBottleneckCalculator free to use?
Yes. The bottleneck calculator, the GPU comparison tool, every CPU and GPU page and every article are free, with no account, no sign-up and no paywall.
Do I need to install anything or run a benchmark?
No. Nothing is installed and nothing is scanned on your PC. You pick your processor, graphics card, resolution and workload from a list, and the estimate is calculated in your browser from published benchmark averages.
How accurate is this bottleneck calculator?
Treat it as a shortlisting tool with a margin of roughly plus or minus 15% against a real benchmark run, and wider in games with unusual engines. It is good at telling you whether a part is being wasted, and poor at predicting an exact frame rate in one specific game.
Where do the CPU and GPU numbers come from?
Each part is given a maximum frame rate derived from average frame rates across a sample of recent AAA games at a high preset, with no ray tracing and no upscaling, normalised to a common test scene. The methodology page explains the formula and lists what the model deliberately ignores.
Who writes the content on this site?
The site is run by PC builders who got tired of bottleneck tools that hand out a percentage with no explanation. Every calculation is published on the methodology page, so nothing rests on you taking our word for it.
How do I report a wrong number?
Use the contact page or email us. If a figure looks wrong and you have benchmark data to back it up, we correct it and credit the correction.