[core] Add CodSpeed C++ benchmarks for protobuf, main loop, and helpers (#14878)

tests/benchmarks/core/bench_application_loop.cpp

@@ -0,0 +1,22 @@
+#include <benchmark/benchmark.h>
+
+#include "esphome/core/application.h"
+
+namespace esphome::benchmarks {
+
+// Benchmark Application::loop() with no registered components.
+// App is initialized by original_setup() in main.cpp (code-generated
+// pre_setup, area/device registration, looping_components_.init).
+// This measures the baseline overhead of the main loop: scheduler,
+// timing, before/after loop tasks, and yield_with_select_.
+static void ApplicationLoop_Empty(benchmark::State &state) {
+  // Set loop interval to 0 so yield_with_select_ returns immediately
+  // instead of sleeping. This benchmarks the loop overhead, not the sleep.
+  App.set_loop_interval(0);
+  for (auto _ : state) {
+    App.loop();
+  }
+}
+BENCHMARK(ApplicationLoop_Empty);
+
+}  // namespace esphome::benchmarks

tests/benchmarks/core/bench_helpers.cpp

@@ -0,0 +1,41 @@
+#include <benchmark/benchmark.h>
+
+#include "esphome/core/helpers.h"
+
+namespace esphome::benchmarks {
+
+// Inner iteration count to amortize CodSpeed instrumentation overhead.
+// Without this, the ~60ns per-iteration valgrind start/stop cost dominates
+// sub-microsecond benchmarks.
+static constexpr int kInnerIterations = 2000;
+
+// --- random_float() ---
+// Ported from ol.yaml:148 "Random Float Benchmark"
+
+static void RandomFloat(benchmark::State &state) {
+  for (auto _ : state) {
+    float result = 0.0f;
+    for (int i = 0; i < kInnerIterations; i++) {
+      result += random_float();
+    }
+    benchmark::DoNotOptimize(result);
+  }
+  state.SetItemsProcessed(state.iterations() * kInnerIterations);
+}
+BENCHMARK(RandomFloat);
+
+// --- random_uint32() ---
+
+static void RandomUint32(benchmark::State &state) {
+  for (auto _ : state) {
+    uint32_t result = 0;
+    for (int i = 0; i < kInnerIterations; i++) {
+      result += random_uint32();
+    }
+    benchmark::DoNotOptimize(result);
+  }
+  state.SetItemsProcessed(state.iterations() * kInnerIterations);
+}
+BENCHMARK(RandomUint32);
+
+}  // namespace esphome::benchmarks

tests/benchmarks/core/bench_logger.cpp

@@ -0,0 +1,54 @@
+#include <benchmark/benchmark.h>
+
+#include "esphome/core/log.h"
+
+namespace esphome::benchmarks {
+
+// Inner iteration count to amortize CodSpeed instrumentation overhead.
+// Without this, the ~60ns per-iteration valgrind start/stop cost dominates
+// sub-microsecond benchmarks.
+static constexpr int kInnerIterations = 2000;
+
+static const char *const TAG = "bench";
+
+// --- Log a message with no format specifiers (fastest path) ---
+
+static void Logger_NoFormat(benchmark::State &state) {
+  for (auto _ : state) {
+    for (int i = 0; i < kInnerIterations; i++) {
+      ESP_LOGW(TAG, "Something happened");
+    }
+  }
+  state.SetItemsProcessed(state.iterations() * kInnerIterations);
+}
+BENCHMARK(Logger_NoFormat);
+
+// --- Log a message with 3 uint32_t format specifiers ---
+
+static void Logger_3Uint32(benchmark::State &state) {
+  uint32_t a = 12345, b = 67890, c = 99999;
+  for (auto _ : state) {
+    for (int i = 0; i < kInnerIterations; i++) {
+      ESP_LOGW(TAG, "Values: %" PRIu32 " %" PRIu32 " %" PRIu32, a, b, c);
+    }
+    benchmark::DoNotOptimize(a);
+  }
+  state.SetItemsProcessed(state.iterations() * kInnerIterations);
+}
+BENCHMARK(Logger_3Uint32);
+
+// --- Log a message with 3 floats (common for sensor values) ---
+
+static void Logger_3Float(benchmark::State &state) {
+  float temp = 23.456f, humidity = 67.89f, pressure = 1013.25f;
+  for (auto _ : state) {
+    for (int i = 0; i < kInnerIterations; i++) {
+      ESP_LOGW(TAG, "Sensor: %.2f %.1f %.2f", temp, humidity, pressure);
+    }
+    benchmark::DoNotOptimize(temp);
+  }
+  state.SetItemsProcessed(state.iterations() * kInnerIterations);
+}
+BENCHMARK(Logger_3Float);
+
+}  // namespace esphome::benchmarks

tests/benchmarks/core/bench_scheduler.cpp

@@ -0,0 +1,133 @@
+#include <benchmark/benchmark.h>
+
+#include "esphome/core/scheduler.h"
+#include "esphome/core/hal.h"
+
+namespace esphome::benchmarks {
+
+// Inner iteration count to amortize CodSpeed instrumentation overhead.
+// Without this, the ~60ns per-iteration valgrind start/stop cost dominates
+// sub-microsecond benchmarks.
+static constexpr int kInnerIterations = 2000;
+
+// --- Scheduler fast path: no work to do ---
+
+static void Scheduler_Call_NoWork(benchmark::State &state) {
+  Scheduler scheduler;
+  uint32_t now = millis();
+
+  for (auto _ : state) {
+    for (int i = 0; i < kInnerIterations; i++) {
+      scheduler.call(now);
+    }
+    benchmark::DoNotOptimize(now);
+  }
+  state.SetItemsProcessed(state.iterations() * kInnerIterations);
+}
+BENCHMARK(Scheduler_Call_NoWork);
+
+// --- Scheduler with timers: call() when timers exist but aren't due ---
+
+static void Scheduler_Call_TimersNotDue(benchmark::State &state) {
+  Scheduler scheduler;
+  Component dummy_component;
+
+  // Add some timeouts far in the future
+  for (int i = 0; i < 10; i++) {
+    scheduler.set_timeout(&dummy_component, static_cast<uint32_t>(i), 1000000, []() {});
+  }
+  scheduler.process_to_add();
+
+  uint32_t now = millis();
+
+  for (auto _ : state) {
+    for (int i = 0; i < kInnerIterations; i++) {
+      scheduler.call(now);
+    }
+    benchmark::DoNotOptimize(now);
+  }
+  state.SetItemsProcessed(state.iterations() * kInnerIterations);
+}
+BENCHMARK(Scheduler_Call_TimersNotDue);
+
+// --- Scheduler with 5 intervals firing every call ---
+
+static void Scheduler_Call_5IntervalsFiring(benchmark::State &state) {
+  Scheduler scheduler;
+  Component dummy_component;
+  int fire_count = 0;
+
+  // Benchmarks the heap-based scheduler dispatch with 5 callbacks firing.
+  // Uses monotonically increasing fake time so intervals reliably fire every call.
+  // USE_BENCHMARK ifdef in component.h disables WarnIfComponentBlockingGuard
+  // (fake now > real millis() would cause underflow in finish()).
+  // interval=0 would cause an infinite loop (reschedules at same now).
+  for (int i = 0; i < 5; i++) {
+    scheduler.set_interval(&dummy_component, static_cast<uint32_t>(i), 1, [&fire_count]() { fire_count++; });
+  }
+  scheduler.process_to_add();
+
+  uint32_t now = millis() + 100;
+
+  for (auto _ : state) {
+    scheduler.call(now);
+    now++;
+    benchmark::DoNotOptimize(fire_count);
+  }
+}
+BENCHMARK(Scheduler_Call_5IntervalsFiring);
+
+// --- Scheduler: set_timeout registration ---
+
+static void Scheduler_SetTimeout(benchmark::State &state) {
+  Scheduler scheduler;
+  Component dummy_component;
+
+  for (auto _ : state) {
+    for (int i = 0; i < kInnerIterations; i++) {
+      scheduler.set_timeout(&dummy_component, static_cast<uint32_t>(i % 5), 1000, []() {});
+    }
+    scheduler.process_to_add();
+    benchmark::DoNotOptimize(scheduler);
+  }
+  state.SetItemsProcessed(state.iterations() * kInnerIterations);
+}
+BENCHMARK(Scheduler_SetTimeout);
+
+// --- Scheduler: set_interval registration ---
+
+static void Scheduler_SetInterval(benchmark::State &state) {
+  Scheduler scheduler;
+  Component dummy_component;
+
+  for (auto _ : state) {
+    for (int i = 0; i < kInnerIterations; i++) {
+      scheduler.set_interval(&dummy_component, static_cast<uint32_t>(i % 5), 1000, []() {});
+    }
+    scheduler.process_to_add();
+    benchmark::DoNotOptimize(scheduler);
+  }
+  state.SetItemsProcessed(state.iterations() * kInnerIterations);
+}
+BENCHMARK(Scheduler_SetInterval);
+
+// --- Scheduler: defer registration (set_timeout with delay=0) ---
+
+static void Scheduler_Defer(benchmark::State &state) {
+  Scheduler scheduler;
+  Component dummy_component;
+
+  // defer() is Component::defer which calls set_timeout(delay=0).
+  // Call set_timeout directly since defer() is protected.
+  for (auto _ : state) {
+    for (int i = 0; i < kInnerIterations; i++) {
+      scheduler.set_timeout(&dummy_component, static_cast<uint32_t>(i % 5), 0, []() {});
+    }
+    scheduler.process_to_add();
+    benchmark::DoNotOptimize(scheduler);
+  }
+  state.SetItemsProcessed(state.iterations() * kInnerIterations);
+}
+BENCHMARK(Scheduler_Defer);
+
+}  // namespace esphome::benchmarks