Short answer: Match your CPU and GPU to the resolution, frame target and workload you actually use. Get a transparent estimate and a practical test plan in seconds.
We checked current manufacturer pages and filled important catalog gaps, including AMD Ryzen X3D, Intel Core Ultra Plus, Radeon RX 9000, GeForce RTX 50 and Intel Arc B-Series models.
Read the verified hardware update →| Scenario | Direction | Tier gap |
|---|
This is a directional estimate, not a guaranteed FPS prediction. Confirm with frame-time and utilization measurements in your actual software.
Context, not guessworkResolution, frame target and workload are part of every result.
Transparent by designSee the model’s limits and learn how to verify the estimate.
Made to be sharedSave a result URL or publish the free responsive widget.
Use the exact parts in your PC, model the resolution and frame target you really use, read the limiting direction, then verify that prediction in the same game or application.
In Windows, open Task Manager and choose the Performance tab. Use the full processor and graphics-card names. On a laptop, select the discrete GPU used by the game, not only the integrated graphics entry.
Check a gaming laptop correctly →Choose the display resolution, intended frame rate and workload you actually run. A 1080p competitive target and a 4K quality target can produce opposite limits from the same hardware.
See how resolution changes the limit →Start with CPU-limited, GPU-limited or balanced. The percentage describes a relative tier gap inside this model. It is not the percentage of FPS lost and it is not a hardware-health score.
Interpret the bottleneck percentage →Repeat one scene, record frame time and utilization, then change one variable. A prediction becomes useful only when the measured response points to the same limiting stage.
Use the PC bottleneck checklist →A balance estimate is not a physical compatibility checker. Use the published calculation and limitations to inspect the model, then build a measured CPU-or-GPU upgrade decision.
The result compares normalized CPU and GPU tier indices after applying the selected resolution, frame target and workload factors. The percentage is the rounded distance between the two adjusted headroom values. It identifies the lower side as the first path to investigate; it does not claim that the same percentage of FPS is lost.
| Input | Why it changes the balance |
|---|---|
| Resolution | Changes the graphics work performed for each rendered frame. |
| Frame target | Changes the time available for CPU frame preparation. |
| Workload | Adds different compute, encoding and memory pressure. |
Every result links to a validation checklist. The complete formula, current coefficients, worked example, limitations and revision history are published in the bottleneck calculator methodology. For focused frame-time and resolution calculations, browse the PC performance calculator collection.
Direct answer: AMD Ryzen 7 9800X3D and NVIDIA GeForce RTX 5090 move from a CPU limit at 1080p to a balanced result at 1440p and a GPU limit at 4K in this 120 FPS gaming scenario. Resolution changes the work per frame, so one universal percentage would hide the useful part of the answer.
| Resolution | Model result | Relative tier gap | Best next check |
|---|---|---|---|
| 1920 × 1080 (Full HD) | CPU-limited | 16% | Busiest CPU-thread time and unused GPU capacity |
| 2560 × 1440 (QHD) | Well balanced | 5% | 1% lows, temperatures and repeatability |
| 3840 × 2160 (4K UHD) | GPU-limited | 30% | GPU frame time after one quality or resolution change |
The gap is a comparison of adjusted headroom, not a promise that the system loses the same percentage of frames. Open the full three-resolution pairing analysis, reproduce the scenario in the calculator, and use the linked checks to test the prediction.
A real diagnosis connects a predicted limit to the way your system behaves. These are the three signals worth checking before you spend.
A CPU can limit a game with modest total usage when one critical thread is saturated. A GPU can sit at 99% and be operating exactly as intended.
Read utilization correctly →1% lows and the frame-time graph expose stalls that an average FPS figure hides.
Understand frame time →Lower resolution, repeat the same scene and compare. The response tells you which stage deserves attention.
Run the 20-minute test →
At 1080p and 240 FPS, the CPU has only a few milliseconds to prepare each frame. At native 4K, the GPU processes four times as many pixels as 1080p. The same PC can be CPU-limited in one scenario and GPU-limited in the next.
Use a repeatable 20-minute test with utilization, frame time, temperatures and controlled setting changes to identify the part limiting your PC.
Read How to Check a PC Bottleneck in 20 Minutes →
Learn what a bottleneck percentage can and cannot show, why results change by workload, and how to verify an estimate with real performance data.
Read What Does a Bottleneck Percentage Mean? →
A budget framework that aligns the display, CPU, GPU, memory, cooling and power supply around one clear target.
Read How to Plan a Balanced Gaming PC Build →Compare how popular combinations shift at 1080p, 1440p and 4K across different frame targets. Open a pairing to see the directional result, the first validation test and detailed guides for both components.
5% relative tier gap in this directional scenario. Validate it with frame-time and utilization data.
Compare three resolutions →20% relative tier gap in this directional scenario. Validate it with frame-time and utilization data.
Compare three resolutions →11% relative tier gap in this directional scenario. Validate it with frame-time and utilization data.
Compare three resolutions →39% relative tier gap in this directional scenario. Validate it with frame-time and utilization data.
Compare three resolutions →9% relative tier gap in this directional scenario. Validate it with frame-time and utilization data.
Compare three resolutions →16% relative tier gap in this directional scenario. Validate it with frame-time and utilization data.
Compare three resolutions →This selection refreshes every 6 hours. Every public guide, tool, component profile, pairing, support page and policy enters the cycle before it begins again.
Use settings that match the performance tier. First test a quality reduction, then decide whether the GPU or an older CPU is the more useful upgrade.
Explore this page →This pairing places more budget discipline on the graphics side. At 1440p, test quality settings and 12 GB VRAM pressure before replacing a capable gaming CPU.
Explore this page →Separate gaming from encode, compile and render work. Rank candidate GPUs by the supported application path as well as resolution performance.
Explore this page →Convert frame budgets, compare pixel load and model CPU-GPU balance with transparent assumptions.
Explore this page →Resolution changes the graphics work performed for each frame. Refresh rate changes how quickly the whole system must deliver it. Define both parts of the display target before judging a CPU-GPU pairing.
Explore this page →Compare average FPS with 1% lows and frame-time data. Learn what each metric shows, why sample length matters and how to diagnose stutter.
Explore this page →A VRAM bottleneck occurs when the working graphics data no longer fits or moves efficiently within available video memory.
Explore this page →The resolution is identical, but a 144 FPS target gives the system less than half the frame time available at 60 FPS.
Explore this page →Understand how 4K shifts performance pressure toward the GPU, when the CPU still matters, and which measurements confirm the limiting stage.
Explore this page →Review what the calculator processes, what the site stores and how third-party services are handled.
Explore this page →Map professional workflows to the stage that actually controls export time, preview smoothness or build speed.
Explore this page →Resolution scale, ray tracing, shadows, volumetrics, reflections and post-processing often affect GPU time more than textures do when VRAM is sufficient.
Explore this page →Good tools state their limits. Our score is a directional estimate designed to help you decide what to measure next.
View all questions →No. Every workload has a slowest stage. It matters when that stage prevents the performance or smoothness you need.
It is a scenario-aware estimate based on normalized component tiers. Real games, settings, cooling and background work still need measurement.
Do not treat a single threshold as a verdict. A small mismatch is normal; investigate when the predicted limit aligns with an unmet target.
The result points to the limiting path and practical checks first. Product choice should follow budget, compatibility and measured need.
