A laptop on battery runs under a different power policy from the same laptop plugged in, so a battery run and a mains run are two measurements, not one measured twice.
- On our validation laptop, an i5-1245U with Iris Xe, the same two builds of the soft shadow stage were run on battery and then plugged in on 3 October 2026. The hard shadow sustained rate read 6.5 to 32.2 fps on battery and 36.5 to 47.3 fps plugged in.
- The spread changed as much as the level: on battery one build's rate moved by a factor of about five between runs; plugged in, every run of both builds stayed inside 36.5 to 47.3 fps.
- SystemCheck's report says at the top when a run was on battery, and asks you to compare it only with other battery runs.
The short answer
Expect a laptop to measure lower and less evenly on battery, and do not compare a battery run with a mains run. How large the gap is depends on the laptop and its power settings, so the honest way to know yours is to measure both, under the same conditions otherwise, more than once each.
What one laptop showed
These figures come from the soft shadow stage's validation on 3 October 2026, recorded in our measurement validation notes. The same two builds of the stage were run interleaved on the same laptop, first on battery (the system reported it discharging, at 54% when the set began) and then plugged in. The stage draws a fixed scene with hard shadows and with soft shadows in turns, and reports a sustained frame rate for each.
- Hard shadows, sustained: 6.5 to 32.2 fps across ten battery runs; 36.5 to 47.3 fps across eight mains runs.
- Soft shadows, sustained: 4.2 to 21.1 fps on battery; 17.5 to 22.6 fps plugged in.
- One build on battery read 6.5 fps in one run and 32.2 fps in another. The spread inside single runs reached 65 points on battery.
- The mains set was run with nothing else open. The battery set's notes record the battery state but not that, so power may not be the only thing that differed. This is one laptop's record, not a rule for every laptop.
What the processor records say, and do not
Our earliest processor runs were also on battery, which the results revealed before anyone chose it. They used an older sampler and a shorter stage than today's Standard run, so their levels are not set against the later mains runs here. Two things from them still hold. First, two three-minute Deep runs on battery settled to 75.6 and 75.5 million operations a second, so a long run on battery can repeat itself. Second, those Deep runs sat about 30% above shorter runs taken earlier the same day on the same battery: the machine itself had changed, from its own power management or from background work, and no single run could have said which.
Ten Standard runs on the same laptop plugged in, on 5 October 2026, gave a sustained processor figure that varied by 2.8% (coefficient of variation). That is the steadiness a second mains run on that laptop can be judged against.
Why the report flags a battery run
When the browser reports the laptop as not charging, the report's run quality section says the run was on battery, with the charge level, and that sustained figures are lower and less repeatable than plugged in. It asks you to compare the run only with other battery runs, or to plug in and run again. The Battery Status API gives a page the charge level and whether the battery is charging, and nothing else: no power policy, no watts, no clock speeds. Chrome and Edge offer it; Firefox and Safari do not, so a run in those browsers cannot be flagged.
Measure your own laptop
Keep everything but the power source the same: the power mode, the screen brightness, the surface, the browser and the programs that are open. Then compare the sustained figures, each with its interval, and look at the shape of the processor curve as well as its level.
For how fast the battery empties under a fixed load, the battery drain test records the charge level against time under a load you choose (idle: the page does nothing but read the level; video: a clip decoded and drawn at 30 frames a second; processor: every core but one kept busy by workers). It needs at least 10 minutes before it shows a rate, because the level moves in 1% steps, and its runtime figure is a straight line from that rate, labelled as one.
- Plugged in: run the stress test twice, with a few minutes' rest between runs.
- On battery, with the same power mode: run it twice more.
- Set each pair side by side. A gap larger than the spread inside each pair is the battery's effect on this laptop.
Terms used here
Questions
- Why is my laptop slower on battery?
- Most laptops apply a lower sustained power limit and a lower boost ceiling on battery, and the operating system's power mode can lower them further. A browser cannot read that policy, but it can measure how much work the machine does under each.
- Should I run a benchmark on battery?
- Only to answer a battery question, such as how a laptop performs away from the desk. For a figure to compare with later runs or with other machines, plug in and record the power mode.