Module 9 — Computer Memory (RAM, ROM, Flash, SRAM, DRAM)

Learning Objectives

After completing this module, you should understand:

• What is memory?
• Why do computers need different memory types?
• SRAM vs DRAM
• RAM vs ROM
• Flash memory
• NAND vs NOR Flash
• EEPROM
• DDR RAM
• Memory Controller
• Embedded memory architecture
• Automotive ECU memory layout

1. What is Memory?

Memory is where a computer stores:

• Instructions
• Variables
• Images
• Audio
• Operating System
• Programs

Without memory,
the CPU has nothing to execute.

Think of memory as a notebook.
CPU = Student
Memory = Notebook
Program = Notes
The student (CPU) continuously reads and writes the notebook (memory).

Memory Hierarchy

The farther from the CPU,
the slower,
but larger,
and usually cheaper.

2. Why Multiple Memory Types?

Question:
Why not build the entire computer using registers?

Because:
Registers are extremely fast but very expensive and consume significant chip area.

Question:
Why not build everything with Flash?

Because:
Flash is non-volatile but much slower and has limited write endurance.

Different memories optimize different trade-offs.

Real-Life Analogy

Imagine your office.

Frequently used items stay closest to you.

3. RAM (Random Access Memory)

RAM is the computer's working memory.
Everything currently running lives in RAM.

Example

Suppose
Windows boots.

The CPU executes code directly from RAM (or caches filled from RAM).

Characteristics

RAM

• Fast
• Read/Write
• Volatile

Volatile means
Power OFF
↓
Everything disappears.

Example

Suppose

int x = 10;

Memory

CPU reads

4. ROM (Read Only Memory)

ROM stores permanent code.

Examples

• BIOS
• UEFI firmware
• Bootloader
• Microcontroller startup code

Characteristics

• Non-volatile
• Data survives power loss
• Traditionally read-only after manufacturing (modern systems often use Flash instead of true ROM)

Example

RAM vs ROM

RAM	ROM
Volatile	Non-volatile
Read/Write	Mostly read during normal operation
Temporary	Permanent firmware
Fast	Generally slower than RAM
Stores running programs	Stores boot firmware

5. SRAM (Static RAM)

SRAM is used inside

• Cache
• Small embedded memories
• High-speed buffers

Why "Static"?

Because
It does not require periodic refreshing while power is applied.
As long as power exists,
the stored value remains stable.

Structure

Each bit typically uses about 6 transistors (6T SRAM cell).

Advantages

• Extremely fast
• No refresh
• Low access latency

Disadvantages

• Large silicon area
• Expensive
• Lower density

Applications

• L1 Cache
• L2 Cache
• L3 Cache
• Register files

6. DRAM (Dynamic RAM)

Main system memory uses DRAM.

Why "Dynamic"?

Each bit is stored as charge in a tiny capacitor.
Charge leaks over time.
Therefore
Memory must be refreshed periodically.

One bit

Only one transistor plus one capacitor (1T1C).
Much denser than SRAM.

Advantages

• Cheap
• High density
• Large capacity

Disadvantages

• Slower
• Requires refresh
• Higher access latency

SRAM vs DRAM

SRAM	DRAM
Very Fast	Slower
No Refresh	Refresh Required
Expensive	Cheap
Low Density	High Density
Cache	Main Memory
~6 transistors/bit	1 transistor + capacitor/bit

Why Cache Uses SRAM

Suppose CPU executes

sum += array[i];

CPU may need data every cycle.
DRAM is too slow.
Therefore
CPU uses SRAM cache.

Why RAM Uses DRAM

Suppose
Computer has
32 GB RAM

Imagine using SRAM.
Cost would be enormous.
Chip size would be impractical.
DRAM provides the required capacity at much lower cost.

7. DDR Memory

Modern RAM is
DDR
Double Data Rate

Meaning
Transfers data on

• Rising clock edge
• Falling clock edge

Older SDR transferred once per clock cycle.

Evolution

Each generation improves

• Bandwidth
• Capacity
• Power efficiency

Embedded Systems

Automotive SoCs commonly use

• DDR3
• DDR4
• LPDDR4
• LPDDR5

depending on performance and power requirements.

8. Flash Memory

Flash stores

• Firmware
• Bootloader
• Application images
• Configuration
• File systems

Characteristics

• Non-volatile
• Electrically erasable
• Slower writes than RAM
• Limited erase/write cycles

Example

ECU

9. NOR Flash

Characteristics

• Random access
• Execute-In-Place (XIP) capable
• Faster reads
• Slower erase/write
• Lower density

Used for

• Boot firmware
• Bootloaders
• Microcontroller code storage

10. NAND Flash

Characteristics

• Block-oriented access
• Very high density
• Faster erase/write than NOR
• Lower cost per bit
• Cannot normally execute code directly

Used for

• SSDs
• eMMC
• UFS
• SD Cards

NOR vs NAND

NOR	NAND
Random Read	Block Read/Write
Execute in Place	Usually copied to RAM before execution
Boot Code	Storage Devices
Lower Density	Higher Density
Faster Reads	Higher Sequential Throughput

11. EEPROM

EEPROM
Electrically Erasable Programmable ROM

Stores

• Device configuration
• Calibration
• MAC address
• Vehicle parameters

Characteristics

• Non-volatile
• Byte-level updates (unlike Flash, which typically erases blocks)

12. Memory Controller

CPU does not directly manage DRAM timing.
Instead

The Memory Controller handles

• Row selection
• Column selection
• Refresh operations
• Timing constraints
• Burst transfers

Modern desktop/server CPUs usually integrate the memory controller on-chip.

13. Embedded Automotive Memory Map

Typical ECU

This layout varies by MCU/SoC, but the concept is similar.

14. ADAS Camera Example

Notice:

• Firmware originates in Flash.
• The kernel and applications execute from RAM.
• Large image buffers reside in DDR.
• Frequently accessed data moves through caches.

15. Why Operating Systems Care About Memory

The OS is responsible for:

• Allocating RAM
• Protecting memory
• Managing virtual memory
• Loading programs
• Managing page caches
• Handling DMA buffers
• Sharing memory between processes
• Freeing memory when processes terminate

Memory management is one of the OS's primary responsibilities.

Interview Questions

Beginner

• What is RAM?
• What is ROM?
• Difference between SRAM and DRAM?
• What is Flash memory?
• What is DDR?

Intermediate

• Why does SRAM not require refresh?
• Why does DRAM require refresh?
• Why is cache built with SRAM?
• Difference between NOR and NAND Flash?
• Why is Flash non-volatile?

Senior

• Explain a complete automotive ECU memory map.
• Why are image buffers stored in DDR?
• Why isn't Flash used as main memory?
• Explain the role of the memory controller.
• How does the bootloader move execution from Flash to RAM?
• Why is cache essential when processing high-resolution camera frames?

Common Mistakes

• ❌ Thinking RAM and DRAM are the same term. DRAM is a technology used to implement most system RAM.
• ❌ Assuming ROM and Flash are identical. Modern systems often replace traditional ROM with Flash, but they are not the same technology.
• ❌ Believing Flash can replace DRAM as working memory. Flash has much higher write latency and limited write endurance.
• ❌ Thinking SRAM is simply "better" than DRAM. SRAM is faster but much more expensive and less dense.

Summary

Modern computers use multiple memory technologies because no single technology provides the best combination of speed, capacity, cost, and persistence. Registers and SRAM provide very high performance close to the CPU. DRAM provides large working memory. Flash stores firmware and persistent data. The operating system coordinates these memory resources so applications can execute efficiently and safely.

Mini Quiz

1. Why does a computer need different memory types?
2. Why is SRAM faster than DRAM?
3. Why does DRAM require refresh?
4. Why is cache implemented using SRAM?
5. Why is main memory implemented using DRAM?
6. Why can't Flash replace DRAM?
7. What is the difference between NOR and NAND Flash?
8. What does the memory controller do?
9. Where does a bootloader typically reside?
10. Where are image buffers stored in an ADAS ECU?