A Step-by-Step Checklist for Diagnosing a PC Bottleneck
Diagnostic plan: Work through a repeatable diagnosis before changing hardware. The checklist keeps the target, baseline, monitoring and one-variable tests in the correct order and stores progress only in this browser.
- 12 ordered checks
- Private browser storage
- Copyable test summary
Name the target before checking boxes
These fields stay in this browser. Specific notes make the completed checklist useful when you repeat the test or ask someone else to review it.
Why the order matters
Target
A limit matters only when it prevents a named outcome.
Baseline
The same scene and capture method reveal normal variance.
Exclusions
Caps, heat, memory setup and background work can imitate a hardware limit.
Response
One changed variable should produce the response predicted by the hypothesis.
Skipping directly to component replacement removes the evidence needed to tell a real hardware ceiling from a configuration problem. Complete the brief first, preserve the baseline and restore it before each new test.
Capture signals from the same run
Total CPU and GPU percentages do not describe the entire performance pipeline. Record the signals that can explain the exact event you are investigating, all from the same time window.
Frame time, average FPS, percentile lows, CPU time and GPU time.
Effective clocks, temperatures, power, limit flags and fan behavior.
RAM, VRAM, paging, storage activity and bus traffic.
Scene, settings, cap, software version, driver and warm-up state.
For productivity and AI work, time the actual stage
Use preview responsiveness, export duration, render time, compile time, model loading or steady inference throughput as appropriate. A gaming percentage cannot replace the stage that controls a real project.
Choose the next branch from the response
| Controlled change | Observed response | Next branch |
|---|---|---|
| Lower resolution or rendering cost | GPU time falls and performance improves | Investigate graphics settings, clocks, VRAM and GPU capacity. |
| Lower resolution or rendering cost | Performance stays flat | Check frame caps, critical CPU threads, engine work and memory. |
| Close background applications | Slow frames or completion time improve | Find the contending process before changing hardware. |
| Compare cold and sustained runs | Clocks fall as temperature or a limit signal rises | Correct cooling or power behavior and repeat the baseline. |
If the suspected cause does not respond, record that result. It narrows the problem and prevents the same unsupported fix from returning later.
Follow the branch supported by your test
Use How to Check a PC Bottleneck when you need the complete timed process. Use PC Bottleneck Diagnostics to select a more specific CPU, GPU, memory, storage or thermal test.
