Problems · How to check
How do I prove my laptop is throttling to the manufacturer?
Support desks dismiss feelings and engage with repeatable measurements. Build three things: a result that repeats under controlled conditions, a comparison against what the same model normally does, and a written record of the conditions each run was made in.
The strongest version of that case is short: this model normally settles at roughly this level, mine settles far below it, here are five runs on a clean machine, on mains, in performance mode, in a cool room.
First, notice when it happens
The causes, ranked
1. Evidence that is not repeatable
most commonHow to confirm it
One screenshot, taken once, with no note of what else was running or how warm the room was.
The fix
Run the test at least three times, spaced out, and save every result. Consistency across runs is what converts an anecdote into a measurement.
2. Uncontrolled conditions
most commonHow to confirm it
On battery, in balanced mode, with updates installing, on a soft surface, in a hot room. Any of these gives support a legitimate reason to dismiss the result.
The fix
Mains power, performance profile, nothing else running, hard surface, cool room, and write those conditions down beside each result. Remove every objection before it is made.
3. No comparison to what normal looks like
commonHow to confirm it
A number on its own means nothing to a support desk, because they have no reference for it either.
The fix
Find reviews and owner reports of the same configuration and cite what it normally sustains. Far below its own cohort is the claim; a bare figure is not.
4. Diagnosing instead of reporting
commonHow to confirm it
Claims about paste, firmware or design flaws invite argument and shift the conversation onto ground you cannot win.
The fix
Report the symptom and the measurement, not the cause. The machine does not sustain what this model normally sustains is their problem to explain.
5. Having modified the machine
occasionalHow to confirm it
Repasting, undervolting or opening the machine gives the maker grounds to refuse the claim outright.
The fix
Make the claim before you modify anything. This is the strongest practical reason to measure first and open the machine only after the warranty conversation has ended.
Measure it instead of guessing
Run the deep test three to five times under stated conditions and keep every result. A set of consistent sustained curves, with the conditions written beside them and a cohort figure to compare against, is exactly the kind of evidence a support desk cannot answer with a suggestion to reinstall the graphics driver.
What software cannot tell you here
- We cannot supply your model's expected figure; reviews and owner reports of the same configuration do.
- We cannot read temperatures, which is fine here: the claim rests on lost performance rather than on a number the maker can contextualise away.
Asked, in people’s own words
- Is thermal throttling covered under warranty?
- Normal throttling is not, because it is designed behaviour. Throttling far beyond what the model normally does can be, since it indicates a defect such as poor paste application, a failed fan or a blocked heatsink. The whole case therefore turns on the comparison against the model's usual behaviour rather than on throttling itself.
- What evidence do manufacturers accept for performance problems?
- Repeatable measurements under stated conditions, ideally several runs, alongside a reference for what the model normally achieves. Vague reports and single screenshots are routinely dismissed. Presenting it the way their own testing would be presented is what gets the case escalated.
- Should I repaste before making a warranty claim?
- No. Opening the machine gives the manufacturer grounds to refuse the claim entirely. Measure, report, and let them do the work while it is their responsibility. Repasting is the option for a machine already out of coverage, or after the claim has been resolved.