Match the source to the question
A source can be accurate and still be wrong for the claim. A specification page tells you what a product supports. It does not prove how fast one game will run. A benchmark can measure one configuration without predicting every PC.
| Question | Start with | What to confirm |
| What does a component support? | Manufacturer specification and platform documentation | Exact model, revision, firmware and power configuration |
| How was frame performance measured? | Capture-tool documentation and a published test method | Scene, settings, duration, warm-up and percentile calculation |
| What bandwidth does an interface provide? | The relevant standards body or primary interface documentation | Generation, lane width, encoding overhead and actual device mode |
| Why did a real workload slow down? | A repeatable capture from the affected system | Frame time, per-stage utilization, clocks, temperatures and limits |
| Can an image or audio file be republished? | The asset owner and the exact license attached to the file | Commercial use, attribution, modification and distribution rights |
Evidence should lead to an action
When a source suggests a likely limit, use the PC bottleneck diagnostic checklist to test one variable at a time. If the evidence supports a CPU or GPU mismatch, compare the components in the CPU and GPU bottleneck calculator. If the cause is memory, storage, temperature or software, follow the matching section of the complete bottleneck guide.