Memory Hierarchy & Cache

Memory Hierarchy in Computer Architecture

The memory hierarchy is an arrangement of different types of computer memory based on their speed, capacity, and cost. It is usually depicted as a pyramid. The goal of the hierarchy is to provide the user with the highest possible speed (at the top) and the highest possible capacity (at the bottom) at a reasonable cost.

1. The Memory Pyramid

1. Registers: Built directly inside the CPU. They hold data currently being processed. Capacity is usually in Bytes (e.g., 64-bit).
2. Cache Memory: Small, extremely fast static RAM (SRAM) placed between the CPU and Main Memory. It holds frequently requested data.
3. Main Memory (Primary Memory): This includes RAM (Random Access Memory) and ROM (Read Only Memory). It is directly accessible by the CPU.
4. Magnetic Disks / Solid State Drives (Secondary Storage): Non-volatile storage used to store data permanently (e.g., HDD, SSD).
5. Optical Disks / Magnetic Tapes (Tertiary Storage): Used for backups and archives (e.g., CDs, DVDs).

2. Main Memory: RAM vs ROM

2.1 RAM (Random Access Memory)

• SRAM (Static RAM): Uses flip-flops. Does not need to be refreshed periodically. Very fast, very expensive. Used for Cache Memory.
• DRAM (Dynamic RAM): Uses capacitors and transistors. Must be refreshed constantly (thousands of times per second) otherwise it loses data. Slower but cheaper. Used for standard Main Memory.

2.2 ROM (Read Only Memory)

• PROM: Programmable ROM (can be written to only once).
• EPROM: Erasable Programmable ROM (erased using Ultraviolet light).
• EEPROM: Electrically Erasable Programmable ROM (erased electrically, used in Flash drives).

3. Cache Memory & Principle of Locality

1. Spatial Locality: If a memory location is accessed, it is highly likely that nearby memory locations will be accessed soon (like reading an Array).
2. Temporal Locality: If a memory location is accessed, it is highly likely it will be accessed again very soon (like a variable inside a Loop).

Cache Mapping Techniques

1. Direct Mapping: Each block of main memory maps to exactly ONE specific cache line. (Fast, but high chance of collision).
2. Associative Mapping: A main memory block can be loaded into ANY line of the cache. (No collisions, but slow to search).
3. Set-Associative Mapping: A compromise. The cache is divided into 'sets'. A memory block can map to ANY line within a specific set.