Computer Memory: RAM, ROM, and Storage

• Memory is storage for instructions and data and is a critical factor in computer performance
• Registers are a type of memory, but with a small capacity
• A memory cell is a one-dimensional matrix with a unique address for data access
• Primary memory uses semiconductors, including RAM and ROM
• Secondary memory includes HDDs, SSDs, flash drives, tapes, disks, and optical drives

Volatile Memory

• Constant power must be supplied to retain data
• Data is lost when power is removed
• Random Access Memory and Cache Memory are examples

Nonvolatile Memory

• Contents are retained without power
• Stored information is recoverable even after power cycling
• Read-Only Memory and Flash Memory are examples

Memory Prefixes

• Kilobyte (KB) is 2 to the power of 10 bytes, or 1,024 bytes
• Megabyte (MB) is 2 to the power of 20 bytes, or 1,048,576 bytes
• Gigabyte (GB) is 2 to the power of 30 bytes, or 1,073,741,824 bytes
• Terabyte (TB) is 2 to the power of 40 bytes, or 1,099,511,627,776 bytes
• Petabyte (PB) is 2 to the power of 50 bytes, or 1,125,899,906,842,624 bytes
• Exabyte (EB) is 2 to the power of 60 bytes, or 1,152,921,504,606,846,976 bytes

Random Access Memory (RAM)

• Storage locations are directly accessible
• The data bus transfers data in or out
• The address bus selects memory locations
• The read signal activates data reading
• The write line activates data writing
• RAM operations include read, write, and chip select

Types of RAM

• Static RAM (SRAM) is about 20 times faster and more expensive than Dynamic RAM (DRAM)
  • D Flip Flops are one-bit static RAM
  • Cache (L1, L2, L3)
• Dynamic RAM (DRAM) uses fewer components for each bit, allowing for high-capacity integrated circuits, such as 4 GB per IC
• Fast Page Mode RAM (FPMRAM) allows quicker data access in the same row by eliminating the need for a row address
• Extended Data Out (EDORAM) transfers blocks of data
• Synchronous DRAM (SDRAM) adds a special interface for synchronization and can run at higher clock speeds than DRAM
• Rambus DRAM (RDRAM) utilizes multiple DRAM banks with a new interface for faster data transfer
• Double Data Rate SDRAM (DDR SDRAM) transfers data on both rising and falling edges of the clock, doubling the data transfer rate

Types of RAM Packaging

• Single In-line Memory Module (SIMM) is a small circuit board with chips on one side
• Dual In-line Memory Module (DIMM) is a circuit board with chips on both sides and a 64-bit data bus
  • SO-DIMMM (Small Outline) is a smaller version of DIMM for notebooks
  • Micro-DIMM supports 144 and 172 pins for laptops
• Rambus In-line Memory Module (RIMM) is similar to DIMM but uses different pin settings to hold 184 or 232 pins

Read-Only Memory (ROM)

• Holds information permanently, even without power
• It is nonvolatile memory

Types of ROM

• Programmable ROM (PROM) can be written to once
• Erasable PROM (EPROM) can be erased with ultraviolet light and reprogrammed, with Flash ROM as a type of EEPROM
• Electrically EPROM (EEPROM) can be erased with specific voltage applications and reprogrammed; NAND Flash Memory and BIOS are examples
  • Flash Memory is nonvolatile with applications like flash drives and is a type of EEPROM that allows multiple memory locations to be written or erased
  • There are two technologies for flash memory: NAND and NOR, with NAND having a smaller access time
  • Most flash memories use NAND technology

Computer Bottlenecking

• Transmission delays occur when the system bandwidth is insufficient for the data being processed
• Many factors can create a bottleneck

Memory Partitioning

• Partitions allocate memory areas for program execution with the partition size defining which processes can run
• The number of partitions determines the number of concurrent processes
• Fixed Partitioning divides memory into static partitions at system generation, loading processes into equal or greater partitions
  • Possibilities include equal or unequal size partitioning
• Dynamic Partitioning uses variable-length partitions, allocating memory exactly as required
  • Issues include small holes in memory and external fragmentation

Virtual Memory

• Stores application data and instructions not currently in use by the CPU, enabling systems to run larger applications
• The hard disk drive (HDD) or solid-state drive (SSD) is extended and incorporates allocations and addresses

Caching

• It represents the memory hierarchy level between the processor and main memory and refers to storage managed to leverage locality of reference, improving execution speed with reduced latency and increased capacity and availability

Types of Cache Memory

• L1 Cache (Primary Cache) is as fast as the processor because it is embedded
• L2 Cache (External Cache) catches recent data from the processor not caught by Level 1
• L3 Cache (Shared Cache) catches recent data across CPU cores and is slower than L1 and L2

Locality of Reference

• Processors tend to access the same memory locations repetitively
  • Temporal Locality means a resource referenced once is likely to be referenced again soon
  • Spatial Locality means a nearby storage location is more likely to be referenced if one location was recently referenced
• Accessing a cache is based on predictions, with the following results:
  • Cache Hit means finding the same piece of data
  • Cache Miss means the data is not in the cache, causing latency and delay

Hardware Architecture

• Computer hardware includes physical parts, categorized as external or internal
• External components (peripherals) control input or output, such as the case and monitor (output) and keyboard and mouse (input)
• Internal components include the hard disk drive, motherboard, and video card

Generation and Advancement of Computer Hardware

• Generations mark technological advancements in computer hardware history
  • 1st Generation (Vacuum Tubes) relied on switching capability like turning on/off a bit
  • 2nd Generation (Transistors) were faster, smaller, reliable, and cheaper and part of the transistor miniaturization process
  • 3rd Generation (Integrated Circuits) Integrated circuits are collections of transistors compacted on a single "semiconductor chip" using Keyboards and mice as new inputs
  • 4th Generation (Microprocessors) saw a significant increase in transistors on a single chip, leading to the invention of the microprocessor
  • 5th Generation (Artificial Intelligence [AI]) moved from VLSI to ULSI Technology, resulting in microprocessors with ten million components and are based on parallel processing with AI software
  • AI is an emerging branch of computer science and interprets how to make computers think like human beings

Microprocessors

• Microprocessors are integrated circuits with thousands of transistors for processing instructions and processes
• Microprocessors are designed to execute logical and computational tasks like arithmetic interprocess, device communication, and input/output management

Microprocessor Components

• Central Processing Unit (CPU) accepts data, processes it, and transfers information
• Arithmetic Logic Unit (ALU) performs arithmetic (addition, subtraction, etc.) and logic (AND, OR, NOT) operations
• Control Unit (CU) controls input/output, generates control signals, and performs instruction execution
• Registers are the fastest memory for holding information
• Main Memory temporarily stores information
• Clock regulates timing and speed, and Clock speed is measured in Hertz (Hz)
• Input/Output Devices are used to communicate, using Ports, or connections that interact with the computer
• Direct Memory Access allows data transfer from storage to memory without the CPU
• Bus: A pathway of data or instructions
  • Data Bus carries data to and from memory
  • Address Bus defines addressable memory locations
  • Control Bus carries control signals to components

Registers

• Registers are the fastest way for a CPU to access data
• Modern architectures operate on moving data between main memory and registers
• Registers hold temporary values (in binary) like data, instructions, addresses, and special codes

Kinds of Registers

• General-Purpose Registers / Accumulator store transient data
• Program Counter Register / Instruction Pointer holds the address of the current instruction
• Memory Address Register holds the address of a memory location
• Memory Data Register holds the data value being stored or retrieved
• Status Registers / Flags track special conditions like arithmetic carry, power failure, and internal errors

CPU Architecture

• CPU architecture defines how a processor operates through hardware/software
  • Von Neumann Architecture uses a single data bus for data and instructions and includes both in memory
  • Harvard Architecture uses separate buses and memories for instructions and data

Instruction Set Architecture

• Instruction Set Architecture allows communication between hardware and software
  • Instruction Set Architecture provides operations codes for commands like ADD, SUB, MULT, DIV etc.
  • Complex Instruction Set Computer executes complex instructions in single operations; x86-x64 processors (Intel) are examples
  • Reduced Instruction Set Computer uses simple instructions divided into multiple instructions; ARM (iPhone) is an example

Microprocessor Operations

• Fetch fetches instructions in parallel
• Decode identifies the opcode and instruction type
• Execute executes parts of each instruction

Parallel Processing

• Parallel processing simultaneously breaks up tasks on multiple microprocessors, reducing processing time, but faces synchronization issues

Pipelining

• Pipelining accumulates and executes computer instructions in an orderly logical pipeline