PC Bottleneck Calculator:
Free CPU & GPU Check

PC Bottleneck Calculator: Check CPU and GPU Balance

An RTX 4070 Ti paired with an i5-10400F gives up a double-digit share of the frames that card can actually produce at 1080p. That mismatch is common, expensive, and avoidable with ten seconds of checking. Most tools compare CPU and GPU benchmark scores and stop there. This one compares measured frame rates at the resolution you play, because that is what decides which part runs out first.

— Bran Deen, PC Hardware Analyst — About this tool

Last updated: October 2026 — Database covers CPUs and GPUs through RTX 5090 and RX 9070 XT

A bottleneck happens when one component maxes out while the other sits idle. At 1080p the processor does more work, so a weak chip holds back even a strong card. At 4K the graphics card carries far more of the load, so an older processor can still pair well with a mid-range card. The right answer depends on your resolution. That is what this calculator works out.

How this calculator works

Most bottleneck tools compare benchmark scores. A score is an abstraction, and two processors with the same score can behave very differently in games. This one works from frames per second instead.

• Processor ceilings come from a CPU-limited test: nine games at 720p driven by an RTX 5090, so the graphics card is never the constraint. The result is the maximum rate at which each chip can prepare frames.
• Graphics ceilings come from the reverse test: twenty-one games at 1080p, 1440p and 4K driven by a Ryzen 7 9800X3D, so the processor is never the constraint.
• The calculator takes the lower of the two. That number is your estimated frame rate. The distance between them is the bottleneck percentage, and the lower side is your limiting component.
• 760 CPUs and 96 GPUs: database updated October 2026, covering everything from the GTX 1060 to the RTX 5090.

ComputerBase measures CPU gaming ceilings at 720p with an RTX 5090 across nine games. A Ryzen 7 9800X3D reaches 182 FPS under those conditions, against 71 for a Ryzen 5 3600. This calculator uses those published figures as each processor's frame-serving limit, then compares them against measured graphics card frame rates.

See the full methodology for how the estimate is calculated and where its accuracy limits are.

What Is a PC Bottleneck?

A bottleneck is a component that limits overall performance because it cannot keep pace with the rest of the system. In gaming the two parts that matter are the processor and the graphics card, and they fail in ways that feel completely different.

A CPU bottleneck means your processor cannot prepare frames fast enough to keep the graphics card busy. You get uneven frame times, stutter in crowded scenes, and graphics card usage sitting well below 100 percent. Strategy and simulation titles expose this first, because they lean hardest on the processor.

A GPU bottleneck means your graphics card cannot render frames fast enough. Frame delivery stays smooth but the rate caps lower. At 1440p and 4K in demanding games this is the ordinary state of affairs, and it is not a fault.

Memory and storage can both throttle a system, though neither behaves like a classic processor or graphics limit. Running 8GB of memory in a modern title forces Windows to page to disk, which feels like stutter rather than a low frame rate. A mechanical drive lengthens load times and makes assets pop in late. Neither shows up as a bottleneck percentage, which is why this calculator reports the CPU and GPU relationship and says so plainly.

One point worth settling early: a bottleneck is not a defect, and no build removes it. Something always limits first. The useful question is never whether you have a bottleneck, but whether the part doing the limiting is the one you would have chosen. Our guide to which bottleneck actually costs you more frames goes through the trade-off in detail.

How Do I Know If I Have a Bottleneck?

Check component usage during a game. One part sitting near 100 percent while the other runs well below it names your limit. The calculator gives you the expected answer; the usage check confirms it on your actual hardware, with your drivers and your settings.

That last step is the fastest test anyone can run, and it needs no software at all. Lowering resolution cuts graphics work and leaves processor work almost untouched. If the frame rate stays flat afterwards, something other than the graphics card is holding it there.

Symptoms are worth knowing too, because they point at the type before you measure anything. Stutter in busy scenes while the average frame rate looks fine is a processor signature. A frame rate that is low but perfectly steady, and drops further the moment you raise settings, is a graphics signature.

For a walkthrough with free tools and what each reading means, see how to check for a bottleneck using free software.

CPU vs GPU Bottleneck: Which One Actually Costs You Frames?

A processor limit costs you more than a graphics limit at the same percentage, and the reason is what each one does to your hardware. When the graphics card is the limit, the card is fully used and the processor has spare capacity. When the processor is the limit, the card is sitting idle. You paid for frames you are not receiving.

The felt difference matters more than the average frame rate. A processor limit hurts 1% lows far more than it hurts the average, which is why a 20 percent processor limit can feel worse in play than a 40 percent graphics limit measuring the same average. Averages hide stutter; 1% lows do not.

There is a second asymmetry worth knowing. A graphics limit has an easy fix, because a faster card raises the ceiling directly and nothing else needs to change. A processor limit often drags a new motherboard and new memory along with it, which turns a card-sized purchase into a platform-sized one.

Most people assume lower is always better and chase zero on both sides. The data says otherwise. A 40 percent graphics-limited result at 1440p describes a healthy system doing exactly what it should. A 15 percent processor-limited result describes one quietly wasting a card you already own.

Two deeper reads: the full comparison of both types, and why being GPU-limited is usually the outcome you want.

Popular CPU and GPU Pairings, Checked at Every Resolution

Every figure below comes from this calculator using the ComputerBase charts listed in the data sources box. Read across a row and you can watch the limiting component change as the resolution rises. Nothing here is estimated from a tier or a multiplier.

Pairing a Ryzen 7 9800X3D with an RTX 5080 returns 162 FPS at 1080p and 67 FPS at 4K, drawn from ComputerBase rasterizer averages across 21 games. The gap between processor and graphics card widens from 11 percent to 63 percent as resolution rises, which is exactly why the limiting component changes.

Two rows deserve a second look. The Ryzen 5 3600 with an RTX 5080 is 56 percent processor-limited at 1080p and only 6 percent graphics-limited at 4K. Same hardware, opposite conclusion. And the Ryzen 5 5500 with an RTX 5070 passes through near-perfect balance at 1440p on its way from one limit to the other.

If one of these cards is on your shortlist, the pairing guides go deeper on each: which processors keep an RTX 5080 fed, the same question for the RTX 5070 Ti, and the RTX 5070 pairing list. Running an older chip instead? The Ryzen 5 3600 has a specific card ceiling worth knowing before you shop.

Does Resolution Change Your Bottleneck?

Yes, and it often reverses which component limits you. Raising resolution gives the graphics card more pixels to draw while the processor's workload barely moves. A processor limit at 1080p can become a graphics limit at 4K with no hardware change at all.

ComputerBase records the RTX 5070 at 115 FPS at 1080p and 45 FPS at 4K across 21 games. A Ryzen 7 7800X3D can feed 162 FPS. At 1080p the processor has little spare capacity. Move to 4K and it has 72 percent, because the card absorbs those extra pixels.

This is why a resolution-blind calculator cannot answer the upgrade question at all. Select the resolution you actually play at, not the one your monitor could theoretically reach.

Three other factors move the result, and they are worth naming because none of them appear in the percentage. Memory speed and configuration change processor throughput materially, and a single-channel or non-XMP setup can cost more frames than a chip upgrade would recover. Game genre decides which side is stressed, with simulation and strategy titles leaning on the processor far more than linear shooters. Upscaling changes the question entirely, because DLSS and FSR render at a lower internal resolution and shift load back toward the processor, which can turn a comfortable graphics limit into a processor one.

Look, if you are shopping on a fixed budget, resolution is the single setting that decides where your money goes. And if you plan to move from 1080p to 1440p soon, run both. You will often find the upgrade you need isn't the one you were about to buy, because the limit moves when the pixel count does.

On what a calculated figure can and cannot tell you, we wrote up the accuracy question honestly, including where this tool's own estimate goes soft.

How to Use Our Bottleneck Calculator

Three fields decide the result: processor, graphics card and resolution. Everything else is optional and feeds the power and cost panels rather than the bottleneck percentage itself.

Variant selection is where most mistakes happen. Super, Ti and Ti Super models differ enough to change the verdict, and picking the wrong one quietly invalidates the result without warning you.

If your part shows a note saying it is scored at tier level, there is no published frame rate for it yet. The answer is still directional, just less precise than a measured pairing. Those entries get replaced as the source charts expand.

What Is a Good Bottleneck Percentage?

Under 25 percent is balanced when the graphics card is the limit, and under 10 percent when the processor is. Two different thresholds, because the same number means opposite things depending on which side is lower. Above 25 percent processor-limited is the one reading worth acting on.

This is where most calculators go wrong. They apply one scale to both cases, so a strong system at 4K gets flagged as critically broken for doing exactly what it should. A Ryzen 7 9800X3D with an RTX 5080 at 1440p reads 42 percent graphics-limited. That pairing is close to the best you can currently buy.

What most guides skip is the 1% low. Average frame rate tells you how fast a game runs; the 1% low tells you how it feels. A processor limit hurts that figure much more than the average, which is the real reason the two thresholds differ rather than any arithmetic convention.

There is a fair argument against percentages altogether. Some builders say any single number flattens too much, and that only your own frame-time graph counts. They have a point about precision.

But a percentage drawn from published frame rates answers the question most people arrive with. That question is whether to buy a processor or a card next. And it answers it before the money is spent.

I have seen conflicting data on where the processor threshold should sit. Some testing puts a meaningful limit as low as 10 percent; other sets do not show it until 20. My read is 25 percent to act and 10 percent to watch, because below 10 the difference falls inside run-to-run variance on most benchmark suites.

Two guides go further on the thresholds themselves: what counts as a good percentage in a gaming build, and the symptoms that mean an upgrade is genuinely overdue.

How to Fix a PC Bottleneck

Try the free fixes first. If you're processor-limited and you haven't switched on your memory profile in BIOS, that's the cheapest frame rate you'll ever buy. Plenty of systems ship with it disabled. Update chipset and graphics drivers, close background applications, and cap your frame rate near your monitor refresh rate to smooth frame delivery.

For a processor limit specifically, lower the settings that lean on it: crowd density, simulation detail, draw distance. Raising resolution also shifts load onto the graphics card, which sounds backwards as a performance fix but works.

Quick Comparison

Quick note on buying: the calculator names your limiting part, so upgrade that one. Replacing the component that already has headroom changes nothing, and it is the most common way people spend money on a problem they do not have.

The full list, ranked by how much each one actually recovers rather than by how often forums mention it, is in seven free fixes for a CPU bottleneck.

Cost-Per-Frame: Which Upgrade Gives the Best Value?

Cost per frame is a component's price divided by the frame rate it delivers at your resolution. It is the most useful value measure available, because it cuts past marketing and shows how much real performance your money buys.

The fastest card is rarely the smartest purchase. Flagship models charge a steep premium for the last ten to fifteen percent of performance. Their cost per frame is poor even though their frame rate is high.

Frame rates from the ComputerBase 1440p rasterizer average, 21 games, October 2026. Price ranges as of October 2026 and they move constantly. Recalculate with the price you can actually pay today.

Notice how flat the first three rows are. Value barely changes from the RTX 4060 up to the RTX 5070, then worsens sharply at the top. That pattern holds across generations, which is what makes the upper mid-range the usual sweet spot rather than any single model.

One thing cost per frame cannot tell you: whether your processor can keep the card fed. Run the pairing through the calculator before you buy. There is no value in a fast card your processor cannot supply, however good its cost per frame looks in isolation.

Bottlenecks in Laptops vs Desktops

You can use this calculator for a laptop, treating the output as optimistic. Pick the closest processor and graphics card names and read the result as a direction rather than a number.

Laptop parts carry desktop names at lower power limits. A mobile graphics chip might run anywhere from 60 to 150 watts depending on the chassis. The same model in a thin laptop and a thick one can differ by thirty percent. Cooling headroom and mains power both matter too, and a machine on battery often runs at a fraction of its plugged-in performance.

The practical approach: run the calculation, then verify with in-game usage on mains power. Laptop limits usually come down to heat and power budget rather than the pairing itself, which means the fix is more often a cooling pad or a power profile than a new component.

Bottleneck Calculator FAQ

What is a PC bottleneck calculator?

How accurate is this bottleneck calculator?

What is a good bottleneck percentage?

Will my CPU bottleneck my GPU?

Can my GPU bottleneck my CPU?

How do I check my CPU and GPU bottleneck?

Does resolution change bottleneck at 1080p, 1440p and 4K?

Should I upgrade my CPU or GPU first?

Is a 10 percent bottleneck okay?

Can I use this bottleneck calculator for laptops?