The median exceeds 120 fps; perfect stability does not
The same session recorded eight freeze events totaling 2.597 seconds. A “144 fps” headline would hide that essential context.
One identified session, not a universal benchmark
The analyzed direct session lasted 16 minutes and 4 seconds. It represents one hardware, software, network and routing context. It does not include a photodiode or high-speed-camera input-to-display measurement.
Why median and P95 do not mean “locked”
A median splits samples in half. P95 describes a percentile bound. Neither proves a constant frame rate. A locked-144 claim would require a tight frametime distribution and no meaningful interruptions.
Eight freezes matter more than an isolated peak
The summary alone cannot assign a cause to each event. Network, encode queue, decoder or scheduling may contribute. Publishing them alongside positive results makes later regressions comparable.
The precise claim this test supports
Lume sustained a median above 120 fps for a 16:04 direct session on the tested setup. It does not promise locked 144 fps everywhere or establish perceived latency.