std::flat_map (C++23)

In a nutshell

A sorted map that swaps the red-black tree for a contiguous array — faster lookups (cache-friendly), more compact memory, but O(n) insertion and erasure.

Header

#include <flat_map>

Core API Cheat Sheet

Operation	Signature	Description
Access element	V& operator[](const K& key)	Access by key; inserts a default value if the key is absent
Find	iterator find(const K& key)	Returns an iterator to the element
Insert	pair<iterator, bool> insert(const value_type&)	Inserts a key-value pair
Erase	size_t erase(const K& key)	Removes the element with the given key
Element count	size_t size() const	Returns the number of elements
Is empty	bool empty() const	Checks whether the container is empty
Clear	void clear()	Removes all elements
Iteration	iterator begin() / end()	Traverses in key order
Lower/upper bound	iterator lower_bound(const K&)	Finds boundaries in sorted order
Contains	bool contains(const K& key) const	(Available since C++23) Checks whether a key exists

Minimal Example

// Standard: C++23
#include <flat_map>
#include <iostream>

int main() {
    std::flat_map<int, const char*> m;
    m[1] = "one";
    m[3] = "three";
    m[2] = "two";

    for (const auto& [k, v] : m) {
        std::cout << k << ": " << v << "\n";
    }
    // 1: one  2: two  3: three  (sorted by key)

    std::cout << std::boolalpha << m.contains(2) << "\n"; // true
}

Embedded Applicability: Medium

• Contiguous storage is CPU-cache-friendly; lookups on small datasets far outperform std::map
• No per-node allocator overhead and less memory fragmentation — a good fit for embedded environments with tight heap space
• Insertion/erasure is O(n), so it is unsuitable for large datasets that change frequently
• Compiler support is still landing (GCC 15+, Clang 20+, MSVC STL not yet); evaluate your toolchain before production use

Compiler Support

GCC	Clang	MSVC
15	20	19.51

See Also

• cppreference: std::flat_map

---

Some content is referenced from cppreference.com, used under the CC-BY-SA 4.0 license