performance of count

[blight19]

the performance of count is not very good,can use record it instead of calc it?

[teemozhu]

func (m ConcurrentMap[K, V]) Count() int {
	count := 0
	for i := 0; i < SHARD_COUNT; i++ {
		shard := m.shards[i]
		shard.RLock()
		count += len(shard.items)
		shard.RUnlock()
	}
	return count
}

The function len(shard.items) only reads the count count at the bottom of the hmap and doesn't go through it; there is no need to count a separate count

[blight19]

func (m ConcurrentMap[K, V]) Count() int {
	count := 0
	for i := 0; i < SHARD_COUNT; i++ {
		shard := m.shards[i]
		shard.RLock()
		count += len(shard.items)
		shard.RUnlock()
	}
	return count
}

The function len(shard.items) only reads the count count at the bottom of the hmap and doesn't go through it; there is no need to count a separate count

there's lock in the function ,the benchmark demonstrate it's necessary to separate count instead of count it

[teemozhu]

func (m ConcurrentMap[K, V]) Count() int {
	count := 0
	for i := 0; i < SHARD_COUNT; i++ {
		shard := m.shards[i]
		shard.RLock()
		count += len(shard.items)
		shard.RUnlock()
	}
	return count
}

The function len(shard.items) only reads the count count at the bottom of the hmap and doesn't go through it; there is no need to count a separate count

there's lock in the function ,the benchmark demonstrate it's necessary to separate count instead of count it

You're right, I read your code, excellent job

[a1eks3y]

These changes can decrease speed of Set() and Delete(). For benchmarks, I think you should use b.RunParallel or smth else to compare performance in multithreading because it's main usage of this library