A bottleneck happens when one part of your PC — usually the CPU or the GPU — limits how fast the rest of the system can go, so you end up paying for performance you never actually see. This calculator compares the CPU and GPU you enter (or already own) and estimates which one is more likely to be the limiting factor at your chosen resolution. It’s built for anyone planning a new build, checking whether a part they already own is worth pairing with an upgrade, or trying to figure out why a “fast” PC still feels sluggish in games.
| CPU + GPU | 1080p | 1440p | 4K |
|---|---|---|---|
| Core i3-12100F + RTX 4060 | GPU-limited, normal (7% CPU headroom) | GPU-limited, normal (39% CPU headroom) | GPU-limited, normal (60%+ CPU headroom) |
| Ryzen 5 7500F + RTX 4060 Ti 8GB | GPU-limited, normal (17% CPU headroom) | GPU-limited, normal (46% CPU headroom) | GPU-limited, normal (60%+ CPU headroom) |
| Core i5-12400F + RTX 4070 | Noticeable CPU bottleneck (15%) | GPU-limited, normal (20% CPU headroom) | GPU-limited, normal (51% CPU headroom) |
| Core Ultra 5 245K + RTX 5070 | Mild CPU bottleneck (11%) | GPU-limited, normal (24% CPU headroom) | GPU-limited, normal (53% CPU headroom) |
| Core Ultra 7 265K + RTX 5070 Ti | Noticeable CPU bottleneck (19%) | GPU-limited, normal (16% CPU headroom) | GPU-limited, normal (48% CPU headroom) |
| Core i9-13900K + RTX 5090 | Severe CPU bottleneck (46%) | Noticeable CPU bottleneck (21%) | GPU-limited, normal (22% CPU headroom) |
Typical AAA game at high settings, the calculator’s defaults. A CPU bottleneck % is the share of the GPU left idle; GPU-limited is normal and shows the CPU’s headroom.
How this is calculated: the formula on our Methodology page. Sources: part specs from each maker’s official product page (our hardware database); scores positioned from publicly reported review averages.
Quick answer
This calculator compares your CPU and GPU’s gaming-performance scores at 1080p, 1440p and 4K and tells you which part limits your frame rate at your chosen resolution, since the CPU matters more at 1080p while the GPU dominates at 4K as pixel count grows. When the CPU is the limit, the % is a CPU bottleneck: the share of your graphics card left idle (under 6% balanced, 6–14% mild, 15–29% noticeable, 30% or more severe). When the GPU is the limit, that is normal for gaming, and the % shows how much spare headroom the CPU has.
What the Result Means
The result names the part that limits your frame rate at the resolution you selected, with one percentage whose meaning depends on that part. If the CPU is the limit, the percentage is a CPU bottleneck: the share of the graphics card’s capacity left unused because the CPU can’t prepare frames fast enough, roughly 100 minus the GPU usage a monitoring overlay would show. Under 6% counts as balanced, 6–14% as mild, 15–29% as noticeable and 30% or more as severe; anything above 60% is shown as 60%+. If the GPU is the limit, the result says GPU-limited, which is the normal and usually desirable state for gaming, and the percentage is the CPU’s spare headroom: a faster CPU reads higher there, and only a faster graphics card would raise frame rates.
It’s important to read this as a directional signal, not a guarantee. Two systems with the same 20% CPU bottleneck reading can behave differently depending on the game, the settings used, and background software running on the machine; the Game type and Graphics settings inputs cover the biggest of those differences.
How This Calculator Works
The calculator does not run a benchmark or execute any game code on your machine. Instead, it uses a gaming-performance score for each part: a GPU’s score stands for its average gaming throughput at high settings, and a CPU’s for the frame rate it can sustain when the graphics card isn’t the limit. These scores are ComputingSkill’s own index, placed from publicly reported third-party review averages rather than computed from core counts or clock speeds, and we don’t benchmark parts ourselves (how the scores work). It then adjusts the comparison based on the resolution you pick, because resolution changes how the workload is split between the two components.
At 1080p, the CPU typically has to do more work per second relative to the GPU, because the GPU can render frames quickly and is often waiting on the CPU to prepare the next one. At 4K, the GPU has far more pixels to push per frame, so it becomes the heavier workload and the CPU’s relative importance drops. The calculator weights its CPU-vs-GPU comparison accordingly for each resolution tier (1080p, 1440p, 4K) rather than giving one fixed answer regardless of resolution. The Game type and Graphics settings inputs shift the balance the same way: esports titles and low settings lean on the CPU, CPU-heavy strategy and simulation games load it much more per frame, and cards with less VRAM than our GPU VRAM Calculator suggests, or a 128-bit memory bus, are scaled down further at 1440p and 4K. Ray tracing, upscaling and frame generation are not modelled.
Worked example: Say you enter a Ryzen 5 5600 (6C/12T, 3.5/4.4 GHz, Zen 3) paired with an RTX 4070 (200W, 5888 CUDA cores, Ada Lovelace) at 1080p. The RTX 4070 sits in the calculator’s mid-range/1440p performance tier, while the Ryzen 5 5600 is a solid but older mainstream CPU. At 1080p the calculator shows a noticeable CPU bottleneck of about 20%: roughly a fifth of the RTX 4070’s capacity goes unused, so GPU usage would sit around 80%. At 1440p the pair is GPU-limited with about 15% CPU headroom, and at 4K it is GPU-limited with about 48% headroom. That is normal, because the RTX 4070 now has to render four times the pixels per frame, which is squarely GPU work.
Important Limitations
This tool is an estimate built from those scores, not a benchmark result and not a substitute for one. Real-world frame rates depend on the specific game engine, the graphics settings and upscaling options in use, driver versions, RAM speed and capacity, storage speed, background processes, thermals, and power limits — none of which this calculator measures directly. Two PCs with the exact same CPU and GPU can post noticeably different real-world results because of these factors.
The calculator also can’t account for game-specific quirks — some titles lean unusually hard on single-thread CPU performance, others are almost purely GPU-limited, and some scale unevenly across CPU core counts. Treat the output as a starting point for deciding whether a part is roughly well-matched, not as a precise performance prediction. If you need exact numbers for a specific game, independent benchmark reviews of that exact title are a better source.
Frequently Asked Questions
What does it mean if my CPU is the bottleneck?
It means your CPU is more likely to be limiting how many frames per second your system can produce, while your GPU has spare capacity it isn’t fully using. This is common at 1080p with a fast GPU paired to an older or lower-core-count CPU, and it usually shows up as a game not scaling much even after a GPU upgrade.
Is a bottleneck always a bad thing?
Not necessarily. Every system has some imbalance between components, and a mild one rarely matters in practice. It becomes worth addressing when the imbalance is large enough that you’re clearly paying for GPU or CPU performance you can’t use at your target resolution and settings.
Does resolution really change which part bottlenecks?
Yes. Higher resolutions shift more of the rendering workload onto the GPU, since it has more pixels to draw per frame, which is why the same CPU/GPU pairing can look CPU-bound at 1080p and GPU-bound at 4K.
Can this calculator tell me my exact FPS?
No — it estimates which component is more likely to be limiting, not a specific frame rate. For an FPS estimate, use the dedicated FPS Calculator, which also works from estimated scores rather than benchmark runs.
Should I upgrade my CPU or GPU based on this result?
The result is a useful data point, but also consider your target resolution, the games you actually play, and your budget. A CPU-bound result at 1080p might matter less if you’re planning to move to 1440p or 4K soon, since the GPU becomes the bigger factor there.