Go Concurrent Map Access Benchmark

Mutex

This benchmark uses a sync.Mutex to lock the map, before accessing it. This makes sure, that only one goroutine can access the map at the same time. For this implementation, you need to write own wrappers, to lock and unlock the map on each access.

Comparison

• Mutex run is 30.58% faster than Sync run.

func BenchmarkMutex_run(b *testing.B) {
	var m = make(map[int]int)
	var mutex sync.Mutex
	var wg sync.WaitGroup

	b.ResetTimer()

	for i := 0; i < b.N; i++ {
		wg.Add(2)

		go func() {
			mutex.Lock()
			m[i] = i
			mutex.Unlock()
			wg.Done()
		}()

		go func() {
			mutex.Lock()
			_ = m[i]
			mutex.Unlock()
			wg.Done()
		}()

		wg.Wait()
	}
}

Mutex run

By Run Count

By CPU Core Count

Sync

This benchmark uses the sync.Map type. This type is thread-safe, so you don't need to write own wrappers to lock and unlock the map on each access.

Comparison

• Sync run is 44.06% slower than Mutex run.

func BenchmarkSync_run(b *testing.B) {
	var m sync.Map
	var wg sync.WaitGroup

	b.ResetTimer()

	for i := 0; i < b.N; i++ {
		wg.Add(2)

		go func() {
			m.Store(i, i)
			wg.Done()
		}()

		go func() {
			m.Load(i)
			wg.Done()
		}()

		wg.Wait()
	}
}

Sync run

By Run Count

By CPU Core Count