The compiler constant-folds ProtoSize::varint(42) to 1 and optimizes
the entire inner loop to a single addition, causing ~62ns jitter-dominated
measurements. Use varying inputs (i & 0x7F for small, 0xFFFF0000 | i for
large) so each call computes a real result.
Sub-microsecond benchmarks are dominated by the ~60ns per-iteration
valgrind start/stop cost in CodSpeed simulation mode. Add kInnerIterations
(1000) inner loops to all fast benchmarks so the actual work dominates.
Move DoNotOptimize calls outside inner loops to prevent artificial overhead.
Also address review feedback:
- Use tokenless CodSpeed (public repo, no CODSPEED_TOKEN needed)
- Fix warning message to show component-specific path
- Fix stray ". :" in error message
- Verify pinned SHA on re-runs to prevent stale checkouts
- Remove BM_ prefix from all benchmark names (unnecessary convention)
- Remove bench_application_loop.cpp (App needs pre_setup/setup which
requires full code generation; will revisit as integration benchmark)
- Fix extra_include_dirs to use os.path.relpath for sibling directories
Add automated benchmarks using Google Benchmark to prevent performance
regressions in the API protobuf encoding/decoding and core loop paths.
Benchmarks cover:
- Protobuf encode: SensorState, BinarySensorState, HelloResponse,
LightState, DeviceInfoResponse (20 nested devices + 20 areas)
- Protobuf decode: HelloRequest, SwitchCommand, LightCommand
- Protobuf calculate_size and full calc+encode send path
- Varint parse/encode/size for various value ranges
- Scheduler call/next_schedule_in with idle and active timers
- Application loop component dispatch and blocking guard overhead
Infrastructure:
- Extract shared build logic from cpp_unit_test.py into test_helpers.py
- Add cpp_benchmark.py mirroring the unit test build pattern
- Support benchmark.yaml per component dir for declaring dependencies
- Add CodSpeed CI workflow triggered on api/core changes
- Fix ProtoMessage protected destructor on host platform
J. Nick Koston