EN ID
Benchmarks

Benchmarks and field validation

These are the current reference comparisons used to judge whether the runtime pays for its footprint: deterministic timing first, then throughput, then memory cost.

Primary runtime comparisons

Wemos D1 mini / ESP8266

Metric Before After
RAM usage 28300 / 80192 29044 / 80192
Fast avg lag 2512 us 0 us
Fast max lag 11054 us 0 us
Fast misses 126 0

ESP32

Metric Before After
RAM usage 22116 / 327680 22820 / 327680
Fast avg lag 2022 us 0 us
Fast max lag 8156 us 0 us
Fast misses 124 0

ESP32

Metric Before After
Sample runs 49 50
Fast avg lag 871 us 0 us
Fast max lag 1780 us 0 us
Fast misses 24 0

ESP32

Metric Before After
HTTP success 23 27
MQTT success 23 27
HTTP rate 88% 87%
MQTT rate 92% 90%

ESP32

Metric Before After
Control runs 99 101
Fast avg lag 241 us 0 us
Fast max lag 3983 us 0 us
Fast misses 9 0

ESP8266 direct AP seismic node

Metric Before After
Sample runs 476 501
Fast avg lag 5393 us 6 us
Fast max lag 21733 us 2378 us
Fast misses 406 1
Successful local sends 5 7

EnvMonitor

ESP32 compare

Before After
Sample runs 49 50 Fast misses 24 0 Max lag 1780 us 0 us

Telemetry + Queue

Gateway workload

Before After
HTTP success 23 27 MQTT success 23 27 HTTP rate 88 % 87 %

IndustrialLoop

ESP32 compare

Before After
Control runs 99 101 Fast misses 9 0 Max lag 3983 us 0 us

Field validation

The seismic direct-AP node remains the strongest field proof: same hardware, same local backend, but far lower lag and nearly zero misses under transport load.

Tradeoff summary
  • Pay a small, bounded memory cost.
  • Get tighter scheduling and safer recovery.
  • Read transport success together with throughput, not as isolated fail counts.
  • Use optional modules only when they are worth it.