Computer Architecture Basics

Information Technology Essentials - Lecture 4

• Computer Architecture: Computers have evolved in structure, size, and performance. However, generally computers have three main components.
• CPU (Central Processing Unit): Also known as a Processor or Microprocessor. It performs arithmetic and logical operations.
• Central Memory: Stores programs to be executed by the computer.
• Peripherals (I/O Devices): Input/output devices used to exchange information to and from the computer.
• Motherboard: Interconnects the CPU, central memory, and peripherals via communication lines (system bus).

Von Neumann Architectural Model

• Represents a computer with a Central Processing Unit, Central Memory Unit, and Input/Output Units.
• Components are connected via System Bus (Control Bus, Address Bus, Data Bus).
• This model illustrates how components interact for basic functionality.

CPU (Recap)

• CPU operations: Executing programs involve performing arithmetic and logical operations.
• CPU Memory: The central memory temporarily stores programs for execution, as well as input and output.
• Input/Output peripherals: Input peripherals receive user input; output peripherals display results to the user.
• Communication: The motherboard connects components through the system bus for communication.
• Modern computers: The Von Neumann architectural model represents general-purpose computers.

Central Processing Unit (CPU)

• Processors (Microprocessors): Complex electronic circuits for fast machine instruction execution.
• Processor Features: Frequency (MHz/GHz), internal memory (cache/registers), word size (8-64 bit), and number of cores.
• CPU Socket: Connects processor to motherboard.
• Cooling: CPUs require cooling (fan) to regulate temperature.

CPU (Structural Components)

• Arithmetic Logic Unit (ALU): Performs arithmetic and logical operations.
• Control Unit (CU): Fetches and decodes instructions.
• Registers: Small, high-speed storage locations, general and special use purpose (like MAR, MBR, PC etc.).
• Cache Memory: Small, fast memory organized in levels (L1, L2, L3).
• Communication Buses: Address, data, and control buses for communication.

Modern CPU Design

• Multicore: Modern CPUs have multiple computing cores on a single chip, for example, Intel Pentium D, Core i3/i5/i9, AMD Ryzen 3 etc.
• Multicore Operation: Each core has its own ALU and CU in most cases, and shared cache.
• CPU Feature Query (Windows): Windows PCs display CPU information in System > About.

Modern CPU Design (GNU/Linux)

• Command: The GNU/Linux lscpu command displays CPU details (architecture, cores, threads, cache size, etc.).

Processor Instruction Set

• Set of Machine Instructions: Defines instructions a processor can decode and execute.

• CISC and RISC: Two major instruction set architectures.

• CISC (Complex Instruction Set Computers): Single instructions to handle multiple low-level operations (e.g., PDP-11).

  • Fewer instructions are required for complex operations.
  • Easier to implement with microprogramming.
  • Typically uses more transistors as instructions have more complex operations.

• RISC (Reduced Instruction Set Computers): One instruction per low-level operation.

  • More instructions are needed for complex operations.
  • More efficient because the instructions are simpler.
  • Generally less expensive as the chip designs are simpler and use fewer transistors.

• x86 architecture: Common in personal computers.

• ARM architecture: Common in mobile devices, embedded systems.

• Specific Example: (x86) MOV, INC, CMP, JE, HLT

• Machine Instructions and Opcodes: The transformation of assembly instructions into machine instructions (OPcodes) handled by assembler programs (like NASM).

• Instruction Byte Sizes: Instruction sets have multiple byte sizes for different instructions (Intel, 1-14 bytes).

• Clock Cycle and Execution: Different instructions may use different clock cycle times for execution.

• Instruction per Clock Cycle: Earlier processors executed only one simple instruction per clock cycle; newer processors (pipelined/superscalar) execute multiple instructions per clock cycle (Instruction-Level Parallelism (ILP)).

Single Processor Systems

• Single CPU: Only one CPU capable of general-purpose tasks.
• Special Processors: Systems may include other special-purpose processors for tasks like device control or on mainframes.
• Example: A computer with only one general-purpose processor, and multiple other special-purpose processors is still considered a single-processor system.
• Graphical Processing Unit (GPU): Specialized for visual and graphical tasks, often used for scientific simulations or cryptanalysis.
• Tensor Processing Unit (TPU): Specialized for high-speed, low-latency machine learning tasks using deep learning algorithms.

Multiprocessor Systems

• Multiple CPUs: Two or more CPUs executing a general-purpose instruction set, share memory and peripherals, etc.
• Applications: Multiprocessor systems, such as those seen in servers, eventually became available on desktops, laptops and smartphones.
• Improvements: Provide higher throughput, response time, turn-around time, greater reliability, compared to single-core systems.

Clustered Systems

• Connected Multiprocessing: Several interconnected computer systems (hosts).
• Functionalities: Hosts are connected via local area networks; share storage and resources; provide high availability.
• Specific applications: Used in high-performance computing (HPC) environments.