1: Introduction to computer architecture

Computer architecture defines the structure and behavior of a computer system as seen by the user, focusing on how hardware components interact to execute instructions efficiently. It bridges the gap between software requirements and underlying hardware capabilities. This discipline dictates critical design choices affecting performance, power consumption, cost, and reliability.

At its core, architecture encompasses the Instruction Set Architecture (ISA), which serves as the contract between software and hardware. The ISA specifies:

• The set of instructions a processor understands
• Data types
• Registers
• Memory addressing modes
• Handling of interrupts

Key components include:

• Central Processing Unit (CPU) for computation
• Memory hierarchies (like RAM and caches) for data storage
• Input/output subsystems (disks, networks) for communication

The CPU executes instructions through a cycle:

1. Fetching instructions from memory
2. Decoding them
3. Executing operations (e.g., arithmetic in the ALU)
4. Storing results

Memory systems provide:

• Volatile retention (RAM)
• Non-volatile retention (storage)
  ...with caches speeding up access to frequently used data. I/O controllers manage peripherals, enabling user interaction and data transfer.

Architecture profoundly impacts performance. Choices like pipeline depth, cache size, and parallelism (e.g., multicore designs) determine execution speed. Designers constantly balance trade-offs—adding features may boost speed but increase complexity, cost, or power usage. For example:

• Larger caches reduce memory latency
• Larger caches consume more silicon area and energy

Historically, the Von Neumann model (explored next) established foundational principles by storing both instructions and data in unified memory. Modern architectures build on this while addressing bottlenecks through innovations like superscalar execution and out-of-order processing.

Understanding architecture is essential for:

• Optimizing software
• Advancing hardware
  It underpins critical concepts like:
• Performance metrics (clock rate, $CPI$)
• $Amdahl’s Law$ for parallel speedup limits
• Benchmark selection for evaluating real-world effectiveness