BENCHMARK_DEMO is a compile-time flag that runs the complete beacon firmware — every screen, the Morse timing engine, the buttons, the audio feedback and the watchdog — on a bench board whose radio module, battery divider and GPS receiver are not connected. It exists so the repository’s benchmark workflows can measure the real firmware behavior on hardware that has only the ESP32, a display and the buttons.

Enable it

Pass the define at build time:

CODE

# PlatformIO
PLATFORMIO_BUILD_FLAGS="-D BENCHMARK_DEMO=1" pio run -e esp32dev

# Arduino IDE / arduino-cli
# Sketch > Add .ZIP Library is not needed; add to build flags:
#   -DBENCHMARK_DEMO=1

Or flip the default in the source:

CODE

#define BENCHMARK_DEMO 1

Warning: never flash a BENCHMARK_DEMO build onto a rescue device. It will not transmit and it will not read the battery.

What it disables

Each subsystem is gated at its entry point, so the rest of the firmware is untouched:

Subsystem	Without the flag	With `BENCHMARK_DEMO=1`
SX1262 radio	`ensureSpiStarted()` + `radio.beginFSK()` on every TX/RX init	Both init functions return success immediately; no SPI traffic to the module
Carrier keying	`txOn()` / `txOff()` call `radio.transmitDirect()` / `radio.standby()`	Both become empty stubs; the Morse engine still waits the exact dot and dash durations, so per-message timing matches a real TX
RSSI sampling	`radio.getRSSI()` every 20 ms (SCAN) or 5 ms (LISTEN)	A virtual noise floor of -120 dBm; the scan and decode pipelines run against it
RX state	`radio.startReceive()` before each dwell	Skipped
Radio sleep	`radio.sleep()` / `radio.standby()` at mode exits	Skipped
Battery	`analogReadMilliVolts()` on GPIO 36 every 5 s, low-battery warning at 3550 mV	`readBatteryMv()` returns 0 and marks the reading invalid; no ADC access, no low-battery blink, no `AEGIS:BATT:low` line
GPS	`gpsSerial.begin()` at boot and `gps.encode()` in every loop	The UART is never opened and NMEA is never parsed; the cached fix stays whatever it was
Deep sleep	`esp_deep_sleep()` after each beacon cycle	Replaced by an idle loop that keeps the remaining-time screen up, so a measurement session can run back to back

The watchdog, buttons, display rendering, CW timing model and serial bridge all keep running — they are what a benchmark measures.

What it does not change

The UI is identical: the same splash, the same screens, the same battery glyph (rendered from the invalid reading), the same RSSI trace at the noise floor.
Morse timing is exact: txOn()/txOff() stubs preserve the delay() structure, so TX cycles take the same wall-clock time as a transmitting unit.
The dashboard and serial protocol behave the same; AEGIS:STATE reports the demo build through its normal fields.
Flash footprint barely changes (roughly 1 percent smaller on core 3.x), because the flag removes call sites rather than features.

Firmware Overview - where each gate sits
Build Configurations - the three practical hardware builds
Serial Debug System - reading AEGIS:STATE on a demo rig

Benchmark Demo Mode

Benchmark Demo Mode

Enable it

What it disables

What it does not change

Related