[bdbarnett]

The cast figures in the runtimes table and the cast guide, re-measured on a Roku TV (OS 15.3.4). Each run was 40 s through tools/run_cast_example.py, with CPython 3.14.6 on Windows, MicroPython 1.30.0-preview on unix and Windows, and the P4 4" panel.

Values are frames a second, as (testris, lv_test_timer).

Runtime and method	Was	Now
Companion channel, python (Linux)	8.2, 9.1	11.0, 9.1 (lv_test_timer not re-run)
Companion channel, python.exe	12.2, 10.6	testris closes at startup (#245), 13.6
Companion channel, micropython (unix)	11.8, 7.4	13.4, 11.3
Companion channel, micropython.exe	8.4, 7.7	7.7, 7.8
Miracast to a Roku, P4 panel	29, 18	-, 29.1 (testris didn't report)

The desktop range in the guide goes from 6 to 10 to 8 to 14. A P4 Companion frame at 480x270 took 266 to 294 ms to show, not about 180 ms, so the guide now says the P4's limit is PNG encoding.

Some figures are unchanged:

• Miracast from CPython to a Roku: re-measured at 30 frames a second, as written.
• HLS: re-confirmed as playing on Linux and Windows CPython and on the P4.
• P4 Companion figure: stays at 2.2, 2.1. The re-run was limited by a weak Wi-Fi link.

These weren't re-measured:

• Miracast to Windows
• the S3 figures
• the guide's 24 and 18 fps for the panel and a headless P4
• the 7.1 to 7.7 s HLS delay