What is the difference between creating a shared_ptr object via constructor and using the make_shared function?

The difference between creating an object via the shared_ptr constructor and the make_shared function lies in the memory allocation mechanism and efficiency.

• shared_ptr constructor: Allocates memory for the object and the control block (reference counters) separately.

  #include <memory>
  
  class MyClass {
  public:
      int value;
      MyClass(int v) : value(v) {}
  };
  
  int main() {
      // Memory for MyClass and the control block are allocated separately
      std::shared_ptr<MyClass> ptr1(new MyClass(10));
      return 0;
  }
  

• make_shared: Allocates memory for the object and the control block in one memory block.

  #include <memory>
  
  class MyClass {
  public:
      int value;
      MyClass(int v) : value(v) {}
  };
  
  int main() {
      // Memory for MyClass and the control block are allocated in one block
      std::shared_ptr<MyClass> ptr2 = std::make_shared<MyClass>(20);
      return 0;
  }
  

Feature	shared_ptr constructor	make_shared
Memory allocation	Object and control block separately	Object and control block in one memory block
Number of memory allocation operations	Two	One
Efficiency	Less efficient	More efficient (less overhead)
Exception safety	Requires caution with new T(...)	Exception-safe
Support for arrays	Yes (with custom deleter)	No (before C++20), yes (with C++20 for shared_ptr<T[]>)

Using make_shared is preferable unless there are specific reasons to use the constructor, as it is more efficient and safer from an exception safety perspective. The constructor may be needed, for example, when creating a shared_ptr from an existing "raw" pointer.