Study Notes on CPU Classifications and Multi-Processor Systems

Introduction

• Welcome message to students.
• Topic of discussion: Classification in engineering, particularly concerning competitive examinations like GATE, ISRO, and UGC NET.
• Emphasis on the importance of studying this topic for exam preparation.

Understanding CPU Functionality

• The Central Processing Unit (CPU) executes programs.
• Important distinction: The CPU does not execute complete programs directly; it executes instructions sequentially.
• Instructions and data are given to the CPU.

Multi-Processor Systems

• Explanation of multi-processor systems.
• Common approaches for categorizing systems with parallel processing capability identified.

Different Structures in Processor Architecture

1. SISD - Single Instruction Single Data Stream

• Characteristics:
  • Execution involves a single instruction stream and a single data stream.
  • Requires one processor, referred to as a uniprocessor system.
  • The control unit generates the instruction stream, which is provided to the processing unit.
  • Data is taken from memory.
  • Executes instructions in sequence; referred to as sequential computer.
  • Example: Intel microprocessor 8085, characterized by a single instruction stream and a single data stream.

2. SIMD - Single Instruction Multiple Data Streams

• Characteristics:
  • Features a single instruction stream but multiple data streams.
  • Multiple processors work on different data streams under the control unit generating a common instruction stream.
  • Best utilized in vector computing.
  • Operations such as addition and multiplication can be performed in parallel using multiple processing units on a single data set.
  • Example: The architecture is applicable for scientific computing involving vector and array operations.

3. MISD - Multiple Instruction Single Data Stream

• Characteristics:
  • Involves multiple instructions but only a single data stream.
  • Multiple control units generate multiple instruction streams, connected to multiple processing units.
  • Example usage is less common and not widely adopted in markets today.
  • The architecture can process multiple instructions for the same data, which limits optimal utilization of processors compared to other arrangements.

4. MIMD - Multiple Instruction Multiple Data Streams

• Characteristics:
  • Comprises multiple instruction streams and multiple data streams.
  • Each processing unit can operate independently on distinct data.
  • Enhanced processing capabilities, as each processor performs various functions independently.
  • Structurally flexible wherein processors share memory access but optimize execution through locally managed processing tasks.

Memory Access Strategies

Shared Memory vs. Distributed Memory

• Shared Memory:
  • All processors share access to a common memory pool.
• Distributed Memory:
  • Each processor has its dedicated local memory.
  • Local memory operations are not equal for different processors, leading to non-uniform memory access (NUMA) characteristics.
  • Clusters: A collection of SMP systems.

Characteristics of Symmetric Multi-Processor Systems (SMP)

• All processors within the system should be comparable in capability and performance.
• Memory access time should be uniform across all processors.
• IO decisions should be shared through a single communication channel.
• Importance of symmetry in processor functionality.

Implications for Competitive Examinations

• Recognition of questions leading to multi-processor architecture can appear in various exams.
• Understanding characteristics, structures, and classifications are crucial in preparing for these assessments.

Summary and Closure

• Classifications discussed in relation to SMP and NUMA systems enhancing understanding of computing structures and their respective applications.
• Encouragement for further discussion and understanding of clustering in processing systems.
• Thank you for attending the discussion.