Chapter 7 - Input/Output and Storage Systems - Chapter 7

Chapter 7 - Input/Output and Storage Systems

Chapter 7 Objectives

  • Understand how I/O systems work, including I/O methods and architectures.
  • Chapters 7.1 – 7.5

7.1 Introduction

  • Data storage and retrieval is one of the primary functions of computer systems.
    • One could easily argue that computers are more useful to us as data storage and retrieval devices than merely as computational machines.
  • All computers have I/O devices connected to them, and for optimal performance, I/O should be minimized.
  • In studying I/O, the focus is on understanding:
    1. Different types of I/O devices.
    2. How these devices operate.

7.2 I/O and Performance

  • Sluggish I/O throughput can negatively affect overall system performance.
    • This effect is particularly pronounced when virtual memory is involved.
  • The fastest processor in the world is ineffective if it spends most of its time waiting for data.
  • A deep understanding of what happens within a computer system enables efficient resource utilization.

7.3 Amdahl’s Law

  • The overall performance of a system results from the interaction of all its components.
  • System performance improvements are most effectively achieved by enhancing the performance of the most heavily used components.
  • Amdahl’s Law is expressed as: S=1(1−f)+fkS = \frac{1}{(1 - f) + \frac{f}{k}} where:
    • SS is the overall speedup,
    • ff is the fraction of work performed by a faster component,
    • kk is the speedup of the faster component.

7.3 Amdahl’s Law (Cont.)

  • Amdahl’s Law provides a method to estimate expected performance improvement from system component upgrades.
  • Example Scenario:
    • Upgrade CPU to be 50% faster for $10,000 or upgrade disk drives to be 150% faster for $7,000.
    • Processes spend 70% of their time running in the CPU and 30% waiting for disk service.
  • Determining which upgrade offers the greater benefit for the lesser cost:
    • CPU upgrade yields a 30% speedup.
    • Disk drive upgrade results in a 22% speedup.
  • Cost analysis:
    • For processor improvement: each 1% improvement costs $333.
    • For disk improvement: each 1% improvement costs $318.
    • Should price/performance be the only concern?

7.4 I/O Architectures

  • I/O is a subsystem of components that moves coded data between external devices and the host system.
  • I/O subsystems include:
    1. Blocks of main memory for I/O functions.
    2. Buses that facilitate data movement into and out of the system.
    3. Control modules located in both the host and peripheral devices.
    4. Interfaces to external components such as keyboards and storage devices.
    5. Cabling or communication links connecting the host system to its peripherals.

7.4 I/O Architectures (Visual Representation)

  • A model I/O configuration includes:
    • Main memory, CPU, motherboard, I/O bus, monitor, I/O modules, device interfaces, device adapters, and control electronics.

7.4 I/O Control Methods

  • I/O can be managed in five general ways:
    1. Programmed I/O:
    • Reserves a register for each I/O device and continually polls each register to detect data arrival.
    1. Interrupt-Driven I/O:
    • Allows CPU to perform other tasks until I/O requests arise.
    1. Memory-Mapped I/O:
    • Shares the memory address space between I/O devices and program memory, simplifying data transfer processes.
    1. Direct Memory Access (DMA):
    • Offloads I/O processing to a specialized chip that manages the transfer details without CPU intervention.
    1. Channel I/O:
    • Utilizes dedicated I/O processors (IOPs) to handle various channels of data flow.

7.4 I/O Architectures (Interrupt Handling)

  • In interrupt-driven I/O, the interrupt controller signals the CPU when any of the interrupt lines are asserted.
  • The status of the interrupt signal is checked at the top of the fetch-decode-execute cycle.
  • A specific code defined by addresses (interrupt vectors) stored in low memory executes when an interrupt occurs.
  • Before executing the interrupt service routine (ISR), the system state is saved and restored post-execution.

7.4 I/O Architectures (Flowchart of Interrupt Process)

  • Steps in the flowchart for handling interrupts:
    1. Start when interrupt signal detected.
    2. Place ISR address into the Program Counter (PC).
    3. Save current variables and registers.
    4. Look up ISR address in the interrupt vector table.
    5. Branch to ISR to execute necessary code.
    6. Restore any saved registers and variables.
    7. Branch back to the top of the fetch-decode-execute cycle to resume processing.

7.4 I/O Architectures (Memory-mapped I/O)

  • Memory-mapped I/O means I/O devices coexist in the same address space as main memory:
    • Each I/O device has a designated block of memory.
    • This technique requires the same instructions to move data to and from both I/O and memory, making it easier for system design.
  • In smaller systems, low-level data transfer details are managed by I/O controllers built into devices.

7.4 I/O Architectures (DMA Configuration)

  • In the DMA configuration, the DMA shares the bus with the CPU but operates with higher priority and can access memory cycles from the CPU.

7.4 I/O Architectures (Channel I/O)

  • Very large systems often implement channel I/O, which employs one or more I/O processors (IOPs) for channel management.
  • Slower devices like terminals and printers are multiplexed into a single faster channel.
    • In IBM mainframes, multiplexed channels are termed 'multiplexor channels' while the faster channels are referred to as 'selector channels.'
  • The IOP governs protocols, issues device commands, translates between storage and memory coding, and can handle file transfers independently of the CPU.

7.4 I/O Architectures (Channel I/O Configuration)

  • A generic illustration of a channel I/O configuration includes:
    • Main memory, CPU, I/O processors (IOPs), various devices, and the local area network.

7.4 I/O Architectures (DMA Circuit)

  • Configuration of a generic DMA showing how DMA circuits connect to a data bus.

Chapter 7 Conclusion

  • I/O systems are crucial to a computer system's overall performance.
  • Amdahl’s Law serves as a quantitative measure of this assertion.
  • I/O systems encompass memory blocks, cabling, control circuitry, interfaces, and media.
  • Control methods for managing I/O include programmed I/O, interrupt-based I/O, DMA, and channel I/O.