Stack vs heap — how program memory is laid out — How Computers Work — CPU, Memory, OS, Processes and Threads

4. Stack vs heap — how program memory is laid out

Basic★ Must knowLesson 4 of 11

A process's virtual memory is split into regions:

text

High addresses
┌──────────────────────┐
│ Stack (per thread)    │  ← method calls, local variables, return addresses; grows down
│        ↓              │
│                       │
│        ↑              │
│ Heap                  │  ← objects created with new; managed by GC in .NET
├──────────────────────┤
│ Static/global data    │  ← static fields, constants
├──────────────────────┤
│ Code (text)           │  ← machine code (JIT output for .NET)
└──────────────────────┘
Low addresses

	Stack	Heap
Stores	Call frames: parameters, locals, return address	Objects (class instances, arrays, strings)
Allocation	Move a pointer — extremely fast	Find space — fast in .NET (bump pointer), but needs cleanup
Freed	Automatically when the method returns	By the garbage collector (.NET/Java) or manually (`free`/`delete` in C/C++)
Size	Small (1 MB per thread by default on Windows)	Large (limited by RAM/address space)
Per	Thread	Process (shared by all threads)
Errors	Stack overflow (deep/infinite recursion)	Out of memory, leaks, fragmentation

csharp

void PlaceOrder()
{
    int qty = 3;                         // local value on the stack (inside PlaceOrder's frame)
    var order = new Order { Qty = qty }; // the reference 'order' is on the stack; the Order object is on the heap
    Save(order);                         // a new frame for Save is pushed on the stack
}                                        // frame popped: qty and the reference disappear; the Order object stays until GC finds it unreachable

Stack overflow example: int F(int n) => F(n + 1); — every call adds a frame until the 1 MB stack is full and the process crashes (in .NET a StackOverflowException cannot be caught).