If you take one idea away from this site, make it this one: a bottleneck is not a property of two parts. It is a property of two parts and a workload. Change the workload and the answer changes, sometimes completely.
It is also the reason any calculator that ignores resolution is guessing, which is one of the things covered in are bottleneck calculators accurate.
What each part is actually doing
Your processor’s job in a game is mostly fixed cost per frame. It runs the game logic, decides what is visible, works out where everything is, and hands the graphics card a list of things to draw. Doing that once costs roughly the same whether the final image is 1920 pixels wide or 3840 pixels wide, because the list of objects is the same list.
Your graphics card’s job scales with pixels. At 4K it is shading a little over four times as many pixels as it is at 1080p, and post-processing, shadows and reflections scale with that too.
So as you raise resolution, one part’s workload multiplies and the other’s barely moves.
What that looks like in practice
Take a fast processor and a fast card. Say the CPU can prepare about 190 frames a second and the card can render 190 frames a second at 1080p, 160 at 1440p, and 108 at 4K.
| Resolution | CPU can feed | GPU can render | You get | Limited by |
|---|---|---|---|---|
| 1080p | 190 | 190 | ~190 | Neither, they are even |
| 1440p | 190 | 160 | ~160 | GPU |
| 4K | 190 | 108 | ~108 | GPU |
Now put a much slower processor in, one that tops out near 95 frames a second:
| Resolution | CPU can feed | GPU can render | You get | Limited by |
|---|---|---|---|---|
| 1080p | 95 | 190 | ~95 | CPU, badly |
| 1440p | 95 | 160 | ~95 | CPU |
| 4K | 95 | 108 | ~95 | CPU, only just |
At 1080p that CPU is wasting half the card. At 4K it costs about 12%. Same two parts, wildly different verdict. To see which of these two situations you are actually in, run the five minute test in is my GPU being held back.
The counterintuitive fix: raising settings is free
If you are CPU limited, raising your graphics settings is free. Not almost free. Free. If your processor caps you at 95 frames and your card is only being asked for 95 frames when it could do 190, then turning shadows up, switching to a higher texture setting or enabling better ambient occlusion costs you nothing until the card’s own limit falls below 95.
People find this hard to believe because they have spent years lowering settings to gain frames. That habit is correct when the card is the limit and actively wasteful when the processor is.
One caveat: texture quality is the exception, because it spends video memory rather than shading time. Raise it past what your card holds and you get stutter instead of free quality — see running out of VRAM is not the same as a bottleneck.
The other counterintuitive one: upscaling will not save you
Upscaling does not help a CPU limited game. DLSS, FSR and XeSS work by rendering fewer pixels and reconstructing the rest. That reduces the graphics card’s workload. If the graphics card was not the thing holding you back, reducing its workload changes nothing, and you have traded image quality for zero frames.
Frame generation is a partial exception: it inserts generated frames after the fact, so it can raise the number on screen even when the processor is the limit. It does not reduce input latency the way real frames do, though, so a CPU limited competitive shooter still feels sluggish — and the frame consistency underneath it does not improve either, which is what 1% lows vs average FPS is about.
What this means for buying decisions
| You play at | Priority | Why |
|---|---|---|
| 1080p high refresh | Processor | The most likely person to be CPU limited |
| 1440p | Balanced | The middle ground where matched builds make the most sense |
| 4K | Graphics card | The processor requirement is modest, and older chips hold up fine |
If you play at 4K, spend the money on the graphics card. The processor requirement at 4K is genuinely modest, and processors that look outdated on paper hold up fine.
If you play at 1080p on a high refresh monitor, the calculus flips. You are the person most likely to be CPU limited, and a faster processor is the upgrade that will show up. Which part to buy is covered in should you upgrade your CPU or your GPU, and how to divide the money on a new build in how to split your budget between CPU and GPU.
Run your own combination through the bottleneck calculator and switch between the three resolution buttons. The gap moving around is the whole lesson.
FAQs
Does resolution affect bottlenecks?
Yes, more than any other single factor. Processor work per frame stays roughly the same as resolution rises, while graphics card work scales with pixel count. The same pair of parts can be badly CPU limited at 1080p and entirely GPU limited at 4K.
Why am I CPU bottlenecked at 1080p but not at 4K?
Because at 1080p your card finishes each frame quickly and then waits on the processor. At 4K the card is doing four times the work, so it becomes the slower of the two and the processor keeps up comfortably.
Does DLSS or FSR help with a CPU bottleneck?
No. Upscaling reduces the graphics card’s workload by rendering fewer pixels. If the card was not your limit, that changes nothing. Frame generation can raise the displayed frame rate, but it does not improve input latency.
Should I raise my settings if I am CPU limited?
Yes. Extra graphics settings are free until your card’s own limit drops below what the processor can feed. You gain visual quality at no cost in frame rate. Textures are the exception, since they spend video memory rather than shading time.
Is a CPU bottleneck at 1080p bad?
Only if it drops you below the frame rate you can actually use. If your processor caps you at 200 FPS on a 144 Hz display, there is nothing to fix.
What resolution is best for a balanced build?
1440p, usually. It sits between the processor-heavy demands of high refresh 1080p and the card-heavy demands of 4K, which is why both parts tend to be well used there.
Can I fix a CPU bottleneck by raising resolution?
In a sense. Raising resolution does not make the processor faster, but it increases the card’s workload until the card becomes the limit. You convert a CPU limit into a GPU limit at a similar frame rate, with a better image.