The honest answer (table)
Why the single number lies
Most bottleneck calculators output something like “67.4% GPU bottleneck” and call it done. That number is almost certainly fake precision. Three reasons:
- It collapses three resolutions into one. The same CPU + GPU pair can be 30% CPU-bound at 1080p, 5% balanced at 1440p, and 25% GPU-bound at 4K. Reporting a single percentage without specifying resolution is meaningless. Our pair pages always show all three.
- It ignores workload.The same parts also report different verdicts for gaming vs streaming vs productivity vs AI. A “15% bottleneck” for gaming might be a 5% bottleneck for productivity.
- The decimal place is theater.Heuristic index normalization is noisy. Anyone reporting “67.4%” instead of “~30% GPU-bound” is signaling authority they don’t have. We report magnitudes in 5% bands and cap at 35%, because going beyond that band is fake precision.
Specific answers to common “is X% bad” questions
Is a 5% bottleneck bad?
No. 5% is inside the noise floor of any bottleneck measurement. Don't make purchasing decisions on a 5% reading. Likely the build is balanced and the calculator's heuristic returned a small number close to zero.
Is a 9% bottleneck bad?
No. Still inside the noise floor for heuristic calculators. 9% reads worse than it is — you'd see no perceptible FPS gain from upgrading. Treat as balanced.
Is a 10% bottleneck bad?
Not really. 10% is at the edge of measurement noise. If the calculator is consistent (same result on multiple runs), free fixes — FPS cap, XMP/EXPO RAM, closing background apps — usually erase the gap. Don't spend money on a 10% reading.
Is a 15% bottleneck bad?
It's the threshold where the number becomes meaningful. At 15%, you're losing real performance — maybe 5–10 FPS in CPU-heavy games. Tune settings first. If you also plan to upgrade the GPU in the next 12 months, plan around it.
Is a 20% bottleneck bad?
Yes, mild. 20% imbalance means one part is clearly limiting the other — you'd see a 10–15% FPS lift from upgrading the bottleneck. Worth budgeting for an upgrade if you have one planned, not worth a panic-buy.
Is a 28% bottleneck bad?
Yes, noticeably. 28% means the limiting part is dragging down the other by a clearly measurable amount. Upgrading the bottleneck usually pencils — expect 15–25% FPS lift in scenarios where the limit matters.
Is a 40% bottleneck bad?
Yes, severe. 40%+ readings usually mean a generational mismatch — e.g., a 6-year-old i5 paired with an RTX 4070, or a Ryzen 5 5600 paired with an RTX 4090. Upgrade the older part; you're wasting the new one's potential.
What matters more than the percentage
Three numbers that tell you more about a build than any bottleneck-percentage figure:
- PREDICTED FPSIn games you actually play, at your actual resolution. A build that says “20% bottleneck” but hits 144 FPS in your favorite game on a 144 Hz monitor is fine.
- 1% LOW FPSThe frame rate during your worst 1% of frames. This is what you experience as stutter. A 100 FPS average with 30 FPS 1% lows feels worse than 80 FPS with 70 FPS 1% lows.
- RESOLUTION DELTAHow much the verdict changes between 1080p and 4K. A build that’s CPU-bound at 1080p and balanced at 4K just needs you to bump to a higher resolution.
// RUN THE NUMBERS
Have a specific build in mind? Run it through the free PC bottleneck calculator and see the verdict at 1080p, 1440p, and 4K — plus predicted FPS in 10 popular games.
Run my build with a real verdict →FAQ
What is a good bottleneck percentage?
0–15% is fine and within measurement noise. 15–25% is worth attention but usually fixable with tuning. 25%+ is worth upgrading the limiting part. Be skeptical of any calculator that reports decimal-place precision like '17.4% bottleneck' — that's fake precision.
Can a bottleneck be 0%?
In theory, only if both parts finish each frame at exactly the same time — basically impossible. Real builds always have a slight imbalance. Calculators report 'balanced' or 0% when the deviation is below a tolerance band (we use 15%), because below that, you can't reliably distinguish the limit from measurement noise.
Why does my bottleneck percentage change with resolution?
The CPU's per-frame work is mostly resolution-independent; the GPU's work scales with pixel count. So as you raise resolution, the GPU's per-frame budget grows while the CPU's stays roughly constant. A build that's CPU-bound at 1080p often becomes balanced or GPU-bound at 4K.
How accurate is a bottleneck percentage?
Most online calculators apply opaque rules and produce numbers that don't correlate well with measured behavior. Look for calculators that anchor their math on real benchmarks (PassMark, Geekbench, TechPowerUp), report a verdict + magnitude instead of a single decimal-place number, and show how the answer changes across resolutions.
// RUN THE NUMBERS
Have a specific build in mind? Run it through the free PC bottleneck calculator and see the verdict at 1080p, 1440p, and 4K — plus predicted FPS in 10 popular games.
Open the bottleneck calculator →