Bottleneck Calculator

PC bottleneck calculator based on measured evidence

A PC bottleneck is the component or limit that determines performance in one specific workload, scene, resolution, settings preset, and frame-rate target. It is not a permanent property of a CPU and GPU pair. This calculator therefore avoids a universal bottleneck percentage and instead evaluates measurements from the game or application you actually use.

How to use measured mode

Run the workload long enough to capture a representative average, then enter average FPS, the target FPS, average GPU utilization, utilization of the busiest CPU thread, and CPU and GPU frame times. Use an uncapped run when possible. The result weighs frame-time evidence most heavily, uses utilization as supporting evidence, and identifies when the system is already close to its target.

target frame-time budget (ms) = 1000 / target FPS

approximate FPS ceiling = 1000 / the slower CPU or GPU frame time

Measured example

At 110 average FPS with a 144 FPS target, 78% GPU utilization, 96% utilization on the busiest CPU thread, 9.1 ms CPU frame time, and 7.0 ms GPU frame time, the CPU side takes longer to prepare a frame. The calculator reports likely CPU-bound behavior for that measured scene. It does not claim the CPU is always the limit in every game or graphics setting.

How to confirm a CPU bottleneck

Lower the resolution or graphics settings while keeping the same scene and test path. If FPS changes very little and GPU utilization remains below full load while one or more CPU threads are heavily used, the workload is probably CPU-limited. Background software, thermal throttling, memory configuration, simulation load, and an FPS cap can produce similar symptoms, so check those before buying hardware.

How to confirm a GPU bottleneck

Reduce resolution, ray tracing, anti-aliasing, or another GPU-heavy setting. A substantial FPS increase with GPU utilization near full load and GPU frame time above CPU frame time indicates a likely GPU limit. A GPU bottleneck is not automatically a fault; it often means the graphics card is being fully used to deliver the selected image quality.

Planning mode uses benchmark ceilings

For a future build, enter a CPU-limited FPS result measured with a sufficiently fast GPU and a GPU-limited FPS result measured at the intended resolution and settings. Use the same game version, test scene, quality preset, upscaling mode, and frame-generation setting. The lower ceiling is the likely limit, and the uncertainty input creates a range around that estimate. Mixing unrelated benchmark suites makes the comparison unreliable.

The uncertainty band is an entered assumption from 0% to 50%, not a measured confidence interval. With CPU and GPU ceilings of 180 and 120 FPS and a 10% band, the lower benchmark is 120 FPS and the planning range is 108-132 FPS. A value outside the supported range is rejected rather than silently changed.

Target status is not proof of a frame cap

Measured FPS at or above the target means the target has been met in that sample. The tool labels a result within 5% below target as near target; it is not the same as reaching it. Neither condition establishes that VSync, a limiter or another cap is active. For example, 200 FPS exceeds a 144 FPS target by 56 FPS; it is not within 5% of that target. Check the actual application and driver settings separately.

Why total CPU utilization can mislead

A game can saturate one important thread while total CPU utilization appears moderate across many cores. Use the busiest relevant thread or per-core view instead of total package utilization. Intel's guide to identifying PC bottlenecks also emphasizes that the limiting component depends on how the computer is being used and that multiple components can contribute.

Frame time is more useful than a generic score

Frame time describes how long each side of the rendering pipeline needs for a frame. The side with the longer recurring frame time usually sets the throughput ceiling. Intel's Graphics Performance Analyzers guide uses frame analysis to isolate CPU- or GPU-bound behavior. NVIDIA's graphics pipeline performance guidance similarly recommends changing rendering workload to locate the limiting pipeline stage.

Limits of this calculator

The tool does not contain a hardware model database, predict exact game FPS from component names, or convert synthetic CPU and GPU scores into a universal loss percentage. It cannot detect shader compilation stutter, storage stalls, insufficient RAM or VRAM, driver problems, thermal or power limits, network latency, game-engine caps, or poor frame pacing. Validate an upgrade decision with repeatable benchmarks from the exact software you care about.

Related PC calculators

Use the GPU calculator for theoretical compute, memory bandwidth, and efficiency. The power supply calculator estimates component load and headroom, while the AI compute calculator models training throughput rather than gaming performance.

Frequently asked questions

Is a bottleneck percentage accurate?

No single percentage applies to every game, scene, resolution, and setting. This page reports the likely limiting side from the evidence entered and shows the underlying values.

What GPU utilization indicates a GPU bottleneck?

Utilization near full load is supporting evidence, especially when GPU frame time is longer than CPU frame time. Utilization alone is not conclusive because caps, telemetry sampling, and engine behavior can affect it.

Does changing resolution change the bottleneck?

Often. Higher resolution usually increases GPU work, while lower resolution can expose a CPU frame-delivery limit. The result can still vary by game and scene.

Should I upgrade the component named by the result?

Not automatically. First repeat the test, check temperatures and clock speeds, remove caps, close background tasks, and compare the measured shortfall with the performance target.