Hardware diagnostic

PC Bottleneck Calculator: CPU vs GPU by Resolution (2026)

By ·

Is your CPU or GPU the frame-rate limit at 1080p, 1440p, or 4K? An honest, resolution-aware estimate built on Tom's Hardware benchmarks.

Local calculation, no signup Runs in your browser Shareable URL

Your CPU

Gaming index from Tom's Hardware 1080p CPU hierarchy.

Your GPU

Raster index from Tom's Hardware GPU hierarchy.

Target resolution

Resolution decides who does more work: the CPU matters most at 1080p, the GPU dominates at 4K.
-
CPU
GPU
Limiter
-
Imbalance
-
GPU VRAM
-
This is an estimate, not a benchmark. It compares aggregated 1080p gaming (CPU) and rasterization (GPU) data from Tom's Hardware, weighted by resolution. Real bottleneck is specific to each game, its engine, your settings, RAM speed, refresh rate, and whether you use DLSS/FSR or frame generation - none of which a calculator can know. Some CPU bottleneck at 1080p is normal for fast GPUs, and being GPU-bound at 1440p/4K is the expected, efficient state - not a fault. Use this as a directional sanity check for a build or upgrade, then cross-check real benchmarks for the games you play.

No bottleneck calculator can tell you your frame rate. What one can tell you is which part is holding that frame rate down, and at which resolution, which is the question people actually have before they spend money.

A bottleneck is one part of the PC failing to keep up with the other. The faster part then sits partly idle. Your frame rate is capped below what the hardware could do. This calculator estimates which part limits you at the resolution you play, using aggregated gaming benchmarks. Pick a processor. Pick a graphics card. Pick a resolution.

Now the honest part, and I would rather put it at the top than bury it three screens down where nobody reads it: this is a directional estimate rather than a benchmark. A real bottleneck depends on the specific game and its engine, your settings, your memory speed, your refresh rate, and whether you run upscaling. None of that is in here. What the tool is genuinely good for is a sanity check before you buy, spotting an obvious mismatch and understanding why the same pair behaves so differently at 1080p versus 4K. Being GPU-bound at 1440p or 4K is not a fault. It is the efficient, intended state, and plenty of people talk themselves into an upgrade they did not need because nobody ever told them that.

The processor scores come from the Tom’s Hardware CPU benchmarks hierarchy (1080p gaming) and the graphics-card scores from the Tom’s Hardware GPU benchmarks hierarchy (rasterized gaming), both current through June 2026. The resolution-shifts-the-limiter principle is well established by reviewers and by Intel’s own developer guidance. I have not re-run any of those benchmarks myself, so this calculator inherits their methodology and their limits along with their numbers.

Worked examples

These are the calculator’s own outputs. Same formula, same browser. No inputs needed.

The tool’s own default state and a very common balanced 1440p gaming build: Ryzen 7 7800X3D + GeForce RTX 5070 Ti, playing at 1440p

  • Inputs: CPU index (1080p gaming) = 85.6 (Ryzen 7 7800X3D). GPU raster index at 1440p = 69.8 (GeForce RTX 5070 Ti; its row is 76.2 / 69.8 / 61.9 / 16GB). Resolution = 1440p, so RES scale = 0.72. VRAM = 16 GB.
  • Result: Headline: ‘GPU-bound (expected at 1440p)’ (neutral/accent colour). Stats row: Limiter = GPU, Imbalance = 41%, GPU VRAM = 16 GB. Bars: CPU - Ryzen 7 7800X3D 86 / 100 (100% wide, dimmed), GPU - GeForce RTX 5070 Ti @ 1440p 50 / 100 (59% wide, highlighted as the limiter). Recommendation text: the GPU sets the frame rate and that is normal at 1440p — the CPU is keeping up with headroom to spare; to gain fps, upgrade the GPU, lower a few settings, or turn on DLSS/FSR.

A classic real upgrade question: an older Alder Lake i5 kept from a 2022 build, paired with a brand-new high-end card, still on a 1080p high-refresh monitor. Core i5-12600K + GeForce RTX 5080 at 1080p

  • Inputs: CPU index = 60.8 (Core i5-12600K — the lowest CPU index in the dataset). GPU raster index at 1080p = 81.9 (GeForce RTX 5080; row 81.9 / 76.7 / 69.8 / 16GB). Resolution = 1080p, so RES scale = 1.00. VRAM = 16 GB.
  • Result: Headline: ‘Moderate CPU bottleneck’ (amber/warn). Stats row: Limiter = CPU, Imbalance = 26%, GPU VRAM = 16 GB. Bars: CPU - Core i5-12600K 61 / 100 (74% wide, highlighted as the limiter), GPU - GeForce RTX 5080 @ 1080p 82 / 100 (100% wide, dimmed). Recommendation text states roughly 26% of the GPU’s potential goes unused at 1080p, and suggests a faster gaming CPU (X3D chips lead), faster RAM, or moving to a higher resolution/settings. Switching the same pair to 4K flips the verdict to ‘GPU-bound (expected at 4K)’ with Limiter = GPU and Imbalance = 43%.

A mainstream all-new mid-range build checked at 1080p: Ryzen 7 7700X + Radeon RX 9070 — the case where the tool says stop worrying and just upgrade for performance

  • Inputs: CPU index = 70.6 (Ryzen 7 7700X). GPU raster index at 1080p = 69.1 (Radeon RX 9070; row 69.1 / 62.1 / 52.1 / 16GB). Resolution = 1080p, so RES scale = 1.00. VRAM = 16 GB.
  • Result: Headline: ‘Well balanced’ (green). Stats row: Limiter = GPU, Imbalance = 2%, GPU VRAM = 16 GB. Bars: CPU - Ryzen 7 7700X 71 / 100 (100% wide), GPU - Radeon RX 9070 @ 1080p 69 / 100 (98% wide, marked as the limiter). Recommendation text: well matched at 1080p, neither part holds the other back much, so spend on whichever you upgrade for performance rather than to fix an imbalance. Note the Limiter stat still reads GPU even though the headline is ‘Well balanced’ — the under-8% rule short-circuits the severity wording but not the limiter readout.

How to read the result

The tool gives you a limiter, CPU or GPU, and an imbalance percentage. Read it as a direction and a rough size. The processor and graphics-card scores come from two different game suites, so the exact number is an approximation, and a result within a few points either way is effectively balanced hardware. Do not chase the last five percent. What the tool reliably tells you is which part is likely holding you back at the resolution you chose, and how that flips as the resolution changes. Catch the obvious mismatch, then confirm with benchmarks for the games you actually play.

Why resolution decides the limiter

The processor does the same work every frame no matter the resolution: game logic, physics, AI, and setting up draw calls do not care how many pixels the screen has. The graphics card does far more work as resolution climbs, because there are far more pixels to shade. 1440p has roughly 1.78 times the pixels of 1080p, and 4K has four times as many. Raising the resolution loads up the graphics card while the processor’s job stays flat.

That is the whole reason the same CPU-and-GPU pair can be CPU-limited at 1080p and clearly GPU-limited at 4K. It is also why reviewers test processors at 1080p, to expose the gaps between chips, and treat 4K as mostly a graphics-card test. If you play at 4K, a mid-range processor paired with a strong card is usually fine; the same processor can hold that card back at 1080p.

Fixing a CPU bottleneck

If the tool flags your processor as the limiter, there are a few real levers. The most direct is a faster gaming chip, and the ones with extra cache (AMD’s X3D parts) lead in most games because gaming leans hard on cache. Faster memory with its profile enabled (EXPO on AMD, XMP on Intel) helps too, since game performance is sensitive to memory latency. The free option is to raise the resolution or settings, which shifts work onto the graphics card and often hides a mild CPU limit entirely. Before blaming the hardware, rule out the cheap causes: background apps, a stale chipset driver, or a power plan that parks cores can all make a CPU limit look worse than it is.

The mismatches that waste money

Two pairings burn money. The first is a top-tier graphics card behind a weak processor at 1080p: the card spends its life waiting on the CPU and you paid for frames you never see. The second is over-buying the processor for a 4K build, where the graphics card is the limiter anyway and the extra CPU headroom does almost nothing. The goal is not zero bottleneck, which is impossible since one part always leads, but a pair that is balanced for your resolution.

If you are planning a build rather than diagnosing one, the PC Builder picks balanced parts to a budget, and the 7800X3D vs 9800X3D upgrade guide shows how much a cache-heavy gaming chip actually moves the needle. For a worked 1440p example, the $1,500 RTX 5060 Ti build is balanced by design.

Assumptions and sources

  • CPU indexes come from the Tom’s Hardware CPU benchmarks hierarchy (1080p gaming) and GPU indexes from the Tom’s Hardware GPU benchmarks hierarchy (rasterized, native resolution), both current through June 2026; parts without consistent figures across the two suites were omitted rather than estimated.
  • The resolution scaling applied to the GPU score (1.00 at 1080p, 0.72 at 1440p, 0.50 at 4K) is a TechFuelHQ approximation of how far a top card’s frame rate falls as pixel count rises, stated openly here rather than hidden in the model.
  • The under-8% “well balanced” band and the severity wording are TechFuelHQ editorial thresholds - the reason the page tells you to read the percentage as direction and rough magnitude, not measurement.

Related guides and tools

Tool

PC Builder: Parts Picker with Compatibility Checks

Interactive PC build picker with socket, chipset, RAM, PSU, and case-clearance checks, plus current Amazon and Newegg …

Article

Ryzen 7 7800X3D vs 9800X3D: Should You Upgrade?

The case for and against upgrading from a Ryzen 7 7800X3D to the 9800X3D: architecture, gaming gap, overclocking, price, …

PC Builds

The $1,500 RTX 5060 Ti Mid-Range 1440p Build for 2026

The best RTX 5060 Ti 1440p build for 2026: Ryzen 7 7700, 32GB DDR5, tested at 78-312 fps in 7 games, repriced live for …

GPU Reviews

RTX 5060 Ti vs RX 9060 XT: The Honest 1440p Showdown for 2026

RTX 5060 Ti vs RX 9060 XT 1440p benchmarks: ~78 fps raster tie across 8 games, 9060 XT ~$270 cheaper, 5060 Ti +14% in …

Gaming

DX12 vs DX11: Which Gives More FPS (and When to Switch)

DX12 vs DX11 for FPS: why DX12 helps when your CPU is the limit, why it can stutter, and a simple rule for picking one …

Frequently asked questions

What is a CPU or GPU bottleneck?
A bottleneck is when one component finishes its work and then waits on the other, so your frame rate is capped below what the faster part could deliver. If the processor cannot prepare frames fast enough, the graphics card sits partly idle and you are CPU-limited; if the graphics card cannot draw frames fast enough, the processor waits and you are GPU-limited. Every system is limited by one or the other at any moment, so a bottleneck is not a defect by itself; it only matters when the two parts are badly mismatched and you are leaving real performance on the table.
Why does the resolution change which part is the bottleneck?
The processor does the same amount of work per frame no matter the resolution: game logic, physics, AI, and draw-call setup do not care how many pixels the screen has. The graphics card does much more work as resolution rises, because there are far more pixels to shade. 1440p has about 1.78 times the pixels of 1080p, and 4K has four times the pixels of 1080p. So as you raise the resolution you pile work onto the graphics card while the processor’s load stays flat. That is why a fast graphics card can be held back by the processor at 1080p, yet be the clear limiter at 4K. It is also why reviewers test processors at 1080p, to expose the differences, and treat 4K as a graphics-card test.
How does this calculator estimate the bottleneck?
It puts both parts on a 0 to 100 gaming-performance scale from Tom’s Hardware, then accounts for resolution. The processor score is treated as roughly resolution-independent, because a chip’s frame-rate ceiling is about the same at any resolution. The graphics-card score is scaled down at higher resolutions to reflect that a card renders fewer frames per second as the pixel count climbs (a top card retains roughly three-quarters of its 1080p frame rate at 1440p and about half at 4K). Whichever adjusted score is lower is the limiter, and the gap between them is the imbalance. These weights are approximations, stated openly so you can judge the result rather than trust a black box.
My result says GPU-bound. Is that bad?
Usually not. Being GPU-bound means the graphics card sets your frame rate and the processor is keeping up with headroom to spare. That is the normal, intended state at 1440p and 4K, and it means your processor is not holding you back. The case worth worrying about is the opposite one: a CPU bottleneck, where a strong graphics card is held back by a weaker processor, most often at 1080p. If you are GPU-bound and want more frames, the levers are a faster graphics card, slightly lower settings, or upscaling such as DLSS or FSR.
How do I fix a CPU bottleneck?
If the tool flags your processor as the limiter, the most direct fix is a faster gaming processor; chips with extra cache, like AMD’s X3D line, lead in most games. Faster system memory and properly enabled memory profiles (EXPO or XMP) also help, because gaming performance is sensitive to memory latency and bandwidth. A no-cost option is to raise the resolution or graphics settings, which shifts work onto the graphics card and often hides a mild CPU limit. Background apps, an outdated chipset driver, or a power plan that parks cores can all make a CPU bottleneck look worse than the hardware itself.
What benchmark data drives the bottleneck model?
The graphics-card scores come from the Tom’s Hardware GPU benchmarks hierarchy, using rasterized gaming at native resolution, normalized so the strongest card equals 100. The processor scores come from the Tom’s Hardware CPU benchmarks hierarchy, using 1080p gaming on a top-tier graphics card so the result reflects processor capability rather than the card. Both are aggregated across many games and updated through June 2026. A few older processors that did not have consistent figures across the two suites were left out on purpose rather than estimated, so every part in the menus is grounded in published benchmarks.
Can I trust the exact percentage?
Treat the percentage as a direction and a rough magnitude, not a precise measurement. The processor and graphics-card numbers come from different game suites, so they are not perfectly comparable, and your real result depends on the specific games you play, your settings, your memory, and your monitor’s refresh rate. A result within a few points either way is effectively balanced. Use this to catch obvious mismatches and to understand how the same pair behaves at different resolutions, then check real benchmarks for the games you care about before spending money.