render_bench.py (from the parallel perf/render-bench work) had its own
grading module, frame_pacing, with its own definition of a missed frame
and its own refresh estimate. The soak already had both in frame_timing,
so the two could have drifted apart on what "late" means.
The bench now gives the display manager a fresh FrameTimingRecorder,
drains it synchronously at the start and end of the graded run, and prints
frame_soak's report with frame_soak's verdict. Its workload is unchanged:
the synthetic strip, --busy load, the shared speed resolver, the
per-frame scrolling announcement. frame_pacing, its tests and its
src.common exports are removed; measure_refresh_hz moves to frame_timing,
where scroll_speeds.py now finds it.
Two ideas from frame_pacing carry over. The bench seeds the recorder with
the idle refresh it measures, so a loop that free-runs (the 827fps bug
the first bench caught) shows as early frames and one stuck at half rate
as late frames, where an estimate taken from their own intervals finds
both self-consistent. And the soak, which has no idle measurement, now
calls a run NOT LOCKED when its refresh estimate beats the configured cap.
The report also gives the rate held while rendering.
Docs: the bench becomes "Without the service" under "Soaking a rig",
keeping its hdpi numbers and the idle-vs-rendering refresh finding.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
There was no way to answer "does this hardware present every frame on
time?" other than watching the panel. `scripts/render_bench.py` drives the
production path -- a real DisplayManager and ScrollHelper, configured
through the same `scroll_config` resolver every ticker uses -- and grades
the run with a new `src.common.frame_pacing`, exiting non-zero when more
than 0.1% of frames slipped a refresh. Exit 2 when the run could not be set
up at all, so a rig that was never measured cannot pass by accident.
A missed frame is defined exactly: an interval that rounds up to at least
one more refresh than its frame hold asked for. The half-refresh rounding
boundary keeps a frame that ran 1ms long on a 10ms refresh out of the
count, because it still presented on the refresh it was meant to.
The verdict that matters more is NOT LOCKED. A loop that never blocked on
vsync reports a perfect zero misses while presenting nothing -- 8ms frames
on a 100Hz panel all land in the one-refresh bucket while running 25% too
fast -- so the report also checks the typical frame is not shorter than the
panel could physically present. That is what caught the first version of
this benchmark announcing its scrolling state once instead of per frame:
the state expires on an inactivity threshold, the dirty-tracking skip then
fires mid-scroll, and the loop free-ran at 827fps.
And the refresh is read back out of the frames rather than taken from an
idle measurement. Driving the matrix is bit-banging on the same machine, so
pushing frames slows the refresh: a Pi 4 on 512x64 measures 100.4Hz idle
and holds 96.3Hz while scrolling. Both are real, and grading against the
idle figure reports a locked loop as 4% slow -- or, once the gap passes
half a refresh, as missing every frame. The gap between the two is itself
worth watching: a rise in it is a render-cost regression even when nothing
is missed.
Measured on hdpi (Pi 4, 512x64, pwm_bits 8), two minutes each:
plain 95.44 fps, 8 missed of 11,449 (0.070%) PASS
--busy 2 95.41 fps, 3 missed of 11,445 (0.026%) PASS
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014RRtqXDCnvnY6EQwhT5CV9
