Processor and graphics card specifications are live

Problems · How to check

Can a bottleneck calculator tell me if my CPU is holding back my GPU?

Not reliably. A calculator that takes a processor name, a graphics card name and a resolution is working from a lookup table of average scores, and it cannot see the one thing that decides the answer, which is what you actually run. The same pair of parts can be processor limited in a strategy game and graphics limited in the next title at the same resolution.

The percentage figure those sites produce is not a measurement of anything. There is no agreed definition of a bottleneck percentage, no standard method behind it, and no way to verify it, so two calculators will happily give you different numbers for the same pair. Treat it as entertainment rather than as data.

First, notice when it happens

From the first minute

A limit that is present from the first second and stays constant is a genuine hardware balance question, and it is the one a calculator is pretending to answer. It shows as a frame rate that will not rise no matter how much you lower the graphics settings.

Only after minutes of use

A machine that starts fast and slows after several minutes does not have a bottleneck problem at all, it has a heat or power problem, and no part swap will fix it. This is worth ruling out first, because a thermally limited laptop looks exactly like an underpowered one if you only measure the last minute.

The causes, ranked

1. The calculator cannot know your workload, which is what decides it

most common

How to confirm it

The same two parts return a confident percentage regardless of whether you play simulation games, competitive shooters, or edit video, because you were never asked.

The fix

Decide the question per workload instead. Run the actual game or application and watch which component is saturated. The answer legitimately differs between two titles on one machine.

2. Settings and resolution shift the limit more than the parts do

most common

How to confirm it

Dropping the resolution raises the frame rate substantially, meaning you were graphics limited. If it barely changes, you were processor limited.

The fix

Use this as the actual test, because it takes two minutes and needs no tools. Halve the resolution and see whether frame rate moves. That single comparison is worth more than any calculator output.

3. Averages hide the thing you actually feel

common

How to confirm it

The average frame rate looks acceptable but the game stutters. Stutter is usually a processor or memory problem showing up in the worst one percent of frames, which an average conceals.

The fix

Look at frame pacing and the slowest frames rather than the mean. A steady 45 frames per second is more pleasant than an unevenly delivered 60, and a calculator has nothing at all to say about pacing.

4. The real limit is thermal, not the balance of parts

common

How to confirm it

Performance is fine for the first several minutes of a session and degrades afterwards, and it recovers after the machine cools down.

The fix

Fix cooling before buying anything. Clear the vents, use a hard surface, consider fresh thermal paste on an older machine. Replacing a processor because a calculator suggested it will not help a machine that was overheating.

What the word bottleneck actually means

In every system there is always something that is the limit, so having a bottleneck is not a defect, it is a property of any machine that is doing work. The useful question is never whether one exists, it is whether the limiting component is the one you would expect for what you are doing, and whether moving it would be worth the money.

A perfectly balanced machine is not the goal and is not achievable anyway, because balance depends on the workload and your workload changes. A machine that is graphics limited in games and processor limited when compiling is a normal machine, not a misconfigured one.

How to actually find out, in about five minutes

Start with the resolution test. Run your real workload, note the frame rate, then drop the resolution substantially and run it again. A large jump means the graphics card was the limit. Almost no change means it was the processor, memory, or the engine itself.

Then check utilisation, carefully. Graphics utilisation near its ceiling with processor utilisation well below suggests graphics limited. The reverse is harder to read, because total processor utilisation across all cores can look low while one core is completely saturated, which is a very common pattern in games. Look at the busiest single core, not the average.

Finally, rule out heat before concluding anything about parts. If the numbers change between minute one and minute ten, you are measuring cooling rather than balance.

Measure it instead of guessing

A SystemCheck deep run loads the processor and the graphics together and compares that against each one measured alone, which is the closest a browser gets to showing you how the two interact under real pressure. It also measures frame pacing rather than only average frame rate, and reports the sustained level separately from the opening burst, so a thermal limit is not mistaken for a hardware balance problem.

What software cannot tell you here

  • Whether a specific game will be processor or graphics limited on your machine. That depends on the engine and the settings, and it has to be measured in the game itself.
  • What a different processor or graphics card would do. Nothing measured on your machine can predict a part you do not own, and any site that claims otherwise is interpolating from averages.
  • Per-core utilisation during a game. A browser cannot see what other applications are doing with the cores.

Asked, in people’s own words

Are bottleneck calculators accurate?
No. They map two part names onto average benchmark scores and cannot see your workload, settings, resolution, memory configuration or cooling, all of which change the answer. The percentage they return has no standard definition behind it, which is why different calculators disagree about identical pairs.
How do I check if my CPU is bottlenecking my GPU for free?
Run your real workload and drop the resolution substantially. If the frame rate rises a lot, the graphics card was the limit. If it barely moves, the processor or memory was. That comparison takes two minutes, uses no tools, and is more trustworthy than any calculator.
Is a 10 percent bottleneck bad?
The figure does not mean anything measurable, so there is no threshold worth acting on. Something is always the limiting component in any working machine. The question worth asking is whether your actual frame rates and frame pacing are acceptable for what you play, which is measurable.
Can a GPU bottleneck damage your PC?
No. Being the limiting component just means one part is waiting on another, which is the ordinary state of every computer. Nothing is stressed beyond its design by it, and no protection is being bypassed.

Name the cause. Three and a half minutes, in the tab you have open.