Basics of CPU

Understanding how the CPU executes instructions helps you reason about performance bottlenecks — whether your system is CPU-bound or I/O-bound changes your optimization strategy.

⚙️ 4: CPU Basics - The Brain in Detail

🎯 Challenge 4: The Restaurant Kitchen

Scenario: You own a restaurant. You need to serve 100 customers per hour.

Option A: Hire one incredibly fast chef who cooks 100 meals/hour Option B: Hire 10 regular chefs, each cooking 10 meals/hour

Option C: Hire 4 chefs, but each works on multiple dishes simultaneously

Question: Which is best? What are the trade-offs?

The Answer: This is exactly the CPU design problem! Modern CPUs use a combination of all three approaches.

🧠 What is a CPU Core?

CPU CORE - The Processing Unit

A core is one independent processing unit:

One core can execute ONE instruction stream at a time.

🔢 Multi-Core CPUs: The Team Approach

EVOLUTION OF CPUS

📅 YEAR 2000: Single Core

Power: 1x Can do: 1 task at a time

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📅 YEAR 2006: Dual Core

Power: 2x (nearly) Can do: 2 tasks simultaneously

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📅 YEAR 2010: Quad Core

Power: 4x (nearly) Can do: 4 tasks simultaneously

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📅 YEAR 2025: Many Cores

High-End Desktop: 32 cores Server CPU: 128+ cores!

🎮 Real-World Example: Gaming

Let's see how cores are used while gaming:

GAME RUNNING ON 8-CORE CPU

Without multiple cores: One core at 370% = Impossible! Game would run at <30 FPS

⏱️ Clock Speed: How Fast the CPU Thinks

CLOCK SPEED (GHz - Gigahertz)

Clock speed = How many cycles per second

1 Hz = 1 cycle per second 1 KHz = 1,000 cycles per second 1 MHz = 1,000,000 cycles per second 1 GHz = 1,000,000,000 cycles per second

Modern CPU: 3.5 GHz = 3,500,000,000 cycles per second!

What happens in one cycle?

Simple instruction (add two numbers): 1 cycle = Fetch, Decode, Execute, Write

Complex instruction (divide): 10-50 cycles

Memory access: 100-300 cycles (cache miss)

🎯 Instruction Execution: The CPU Pipeline

How does a CPU execute instructions?

THE 4-STAGE PIPELINE

Classic pipeline:

Stage 1: FETCH ├─ Get instruction from memory └─ "Retrieve ADD instruction"

Stage 2: DECODE ├─ Figure out what instruction means └─ "ADD: Add two numbers"

Stage 3: EXECUTE ├─ Perform the operation └─ "5 + 3 = 8"

Stage 4: WRITE BACK ├─ Write result back └─ "Register now contains 8"

🚨 Common Misconception: "Higher GHz Always Faster"

You might think: "5 GHz CPU must be faster than 3 GHz!"

The Reality: It's more complex!

❌ NAIVE COMPARISON:

CPU A: 5.0 GHz, 4 cores CPU B: 3.5 GHz, 8 cores

Your assumption: A is 43% faster!

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

✅ REAL-WORLD RESULTS:

Single-threaded task (video game main thread):

├─ CPU A: 100 FPS ✓ (Winner!)

└─ CPU B: 70 FPS

Multi-threaded task (video rendering):

├─ CPU A: 4 min

└─ CPU B: 2.5 min ✓ (Winner!)

Why?────────────────────────────────────

Single-threaded: Only one core used

├─ Higher GHz wins

└─ CPU A's 5 GHz beats B's 3.5 GHz

Multi-threaded: All cores used

├─ More cores win

└─ CPU B's 8 cores beat A's 4 cores

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

OTHER FACTORS THAT MATTER:

Architecture efficiency:

├─ Instructions per cycle (IPC)

├─ Some CPUs do more per clock

└─ Example: Apple M3 beats Intel at same GHz!

Cache size:

├─ Larger cache = fewer RAM accesses

└─ Can matter more than 0.5 GHz!

Memory speed:

├─ CPU waiting for RAM = wasted cycles

└─ Fast RAM helps more than high GHz

Power efficiency:

├─ High GHz = high power = thermal throttling

└─ Sustained 4 GHz > burst 5 GHz that throttles

🎮 Decision Game: Choose Your CPU

Scenario: Pick the best CPU for each task:

CPU Options:

A. 4 cores, 5.5 GHz, 16 MB cache, $300

B. 8 cores, 4.0 GHz, 32 MB cache, $350

C. 16 cores, 3.0 GHz, 64 MB cache, $500

Tasks:

1. Gaming (mostly single-threaded)

2. Video editing (multi-threaded)

3. 3D rendering (highly parallel)

4. Office work (light multitasking)

5. Software development (compiling code)

Think about each one...

ANSWERS:

1. Gaming → CPU A

   Why: High single-thread performance

   5.5 GHz handles main game thread best

2. Video editing → CPU B

   Why: Good balance

   8 cores for timeline processing

   4 GHz still decent for playback

3. 3D rendering → CPU C

   Why: Maximum parallelism

   16 cores render 16 pixels simultaneously

   3 GHz sufficient per thread

4. Office work → CPU A or B

   Why: Overkill for Office!

   Even CPU A is excessive

   (Budget option would work fine)

5. Software development → CPU B

   Why: Balanced

   Compiling uses all cores

   High clock helps IDE responsiveness

   32MB cache helps with large projects

Key Takeaways

1. The CPU executes instructions in a fetch-decode-execute cycle — understanding this helps you reason about computational bottlenecks
2. Clock speed (GHz) measures how many cycles per second — but more cycles doesn't always mean faster execution due to instruction complexity
3. Multi-core CPUs enable parallel processing — but software must be designed for concurrency to take advantage of multiple cores
4. CPU-bound vs I/O-bound workloads need different optimization strategies — CPU-bound benefits from faster processors, I/O-bound from async/non-blocking designs