Shared Memory Multiprocessors — Parallel Processors

Imagine a group of friends working on a school project.
They sit around the same table, using one notebook to write down ideas.
Each friend adds or reads from the same notebook whenever needed.

This is exactly how shared memory multiprocessors work!
All processors (the “friends”) share the same main memory (the “notebook”), and they can access it whenever they need data or instructions.

---

⚙️ What Is a Shared Memory Multiprocessor?

A shared memory multiprocessor is a computer system that has two or more processors connected to a single shared memory space.

Each processor can directly read from or write to this common memory.
Because they all share the same memory, communication between processors becomes fast and simple — they just read or update shared variables.

---

💡 Simple Analogy: One Kitchen, Many Cooks

Think of a restaurant kitchen with multiple chefs (processors) cooking different dishes.
They all have access to the same fridge and pantry (shared memory).
When one chef takes tomatoes from the fridge, others immediately see that fewer tomatoes are left.

That’s how shared memory works — every processor can see the latest updates made by others.

---

🧩 Basic Structure of Shared Memory Multiprocessor

Let’s look at how it’s organized:

1. Multiple Processors (CPUs):
   Each has its own cache and registers for quick access to data.
2. Shared Main Memory:
   A common memory that all processors can access.
3. System Bus or Interconnection Network:
   Connects all processors to the shared memory, managing communication between them.
4. Input/Output Devices:
   Shared among processors for user interaction and data transfer.

---

🧭 Diagram: Shared Memory Multiprocessor

         +-----------------------------------------+
         |           Shared Memory System          |
         +-----------------------------------------+
             |               |               |
      +--------------+ +--------------+ +--------------+
      | Processor 1  | | Processor 2  | | Processor 3  |
      | (with Cache) | | (with Cache) | | (with Cache) |
      +--------------+ +--------------+ +--------------+
             |               |               |
          -----------------------------------------
                       System Bus

Each processor can independently execute tasks but uses the shared memory to coordinate or exchange data.

---

⚙️ How It Works (Step-by-Step)

1. Task Assignment:
   The main program is divided into smaller parts or threads.
   Each processor works on one part.
2. Memory Access:
   When a processor needs data, it fetches it from the shared memory.
3. Communication:
   Processors share intermediate results by writing them to shared memory.
4. Synchronization:
   They use special control mechanisms (like locks or semaphores) to make sure two processors don’t overwrite each other’s data.

---

🔄 Types of Shared Memory Multiprocessors

There are mainly two architectures based on how memory is accessed:

1. Uniform Memory Access (UMA)

• All processors take the same amount of time to access any memory location.
• The memory is shared evenly through a single bus or interconnection network.
• Common in small-scale systems.

🧩 Analogy:
Like a kitchen where every chef can reach the fridge in the same time — it’s at the center of the room.

---

2. Non-Uniform Memory Access (NUMA)

• Each processor has its own local memory, but can also access other processors’ memory.
• Accessing local memory is faster than remote memory.
• Used in large-scale systems to reduce memory traffic.

🧩 Analogy:
Imagine a big restaurant with multiple kitchens.
Each kitchen has its own fridge (local memory), but chefs can still borrow ingredients from other kitchens — it just takes a bit longer.

---

⚡ Advantages of Shared Memory Multiprocessors

Advantage	Explanation
Fast Communication	Processors share data directly via memory instead of complex message passing.
Easy to Program	Programmers can use shared variables for coordination instead of complicated network code.
Efficient Resource Use	All processors can access the same memory and I/O devices.
Scalability (in moderate systems)	Adding more processors can increase performance up to a limit.

---

⚠️ Challenges and Limitations

Challenge	Explanation
Memory Contention	When multiple processors try to access memory at the same time, it causes delays.
Cache Coherency Problems	Processors may have outdated copies of data in their local caches.
Synchronization Overhead	Managing access to shared data safely (using locks) can slow performance.
Limited Scalability	The shared bus becomes a bottleneck when too many processors are added.

---

🔄 Cache Coherence Problem (Simplified Example)

Let’s say Processor 1 updates a variable X = 10 in memory.
Processor 2 has an old copy of X = 5 in its cache.
If Processor 2 keeps using its old value, the system becomes inconsistent.

To fix this, we use cache coherence protocols (like MESI) that make sure every processor sees the most recent value of shared data.

---

🧮 Synchronization Tools

To avoid data conflicts, processors use synchronization mechanisms such as:

• Locks / Semaphores: Prevent more than one processor from accessing shared data at the same time.
• Barriers: Make all processors wait until everyone reaches a certain point before moving on.
• Monitors: Manage access to shared resources automatically.

These tools help maintain order when multiple processors work together.

---

🧠 Real-World Examples

• Multicore CPUs in laptops and desktops — each core shares the same main memory.
• Servers handling multiple users or web requests at the same time.
• Parallel programming models like OpenMP — built for shared memory systems.

---

About the Author

examhopeinfo@gmail.com

Administrator

Visit Website View All Posts