Assignment SF

Scenario:

Traditionally, computers were seen as machines that handle tasks one step at a time. Most programming languages are designed this way, where programmers write algorithms as a sequence of instructions. The processor then follows these steps, one by one, carrying out actions in a particular order: first, it fetches the instruction, then fetches any needed data, performs the operation, and finally stores the result.

However, this view isn’t completely accurate. At a lower level, computers often perform multiple tasks at the same time. For example, instruction pipelining allows some stages of one instruction (like fetching) to overlap with stages of another, which has been a common technique for years. Both of these are examples of parallel operations happening within the computer.

The concept of parallelism is expanded further in superscalar architecture, where a single processor has multiple execution units. This allows it to handle several instructions from the same program at once, enhancing the computer's overall performance.

As technology has advanced and hardware has become more affordable, computer designers have increasingly looked for ways to use parallelism to boost performance and, in some cases, improve reliability. In a setup called Symmetric Multiprocessing (SMP), multiple processors share the same memory, which brings up issues like cache coherence—ensuring data is consistent across all processor caches.

Clusters, where multiple computers work together, have also become popular for handling larger workloads that one SMP system can’t manage alone. Another approach to using multiple processors is Nonuniform Memory Access (NUMA), which is still relatively new and not as widely used but is considered as an alternative to SMP or clustering.

1. Introduction

  • Briefly introduce the concept of parallel processing and its importance in modern computing.

  • State the purpose of the assignment: to explore various parallel processor architectures and critically analyze the MIMD category.

  • Mention the outcome of the assignment, such as gaining an understanding of different parallel processing models and their practical applications.

2. Taxonomy of Parallel Processor Architectures

/

  • Define and describe the four types of architectures:

    • SISD (Single Instruction, Single Data Stream): A single processor performing one operation on one data stream at a time. Mention it as the traditional, sequential computing model.

    • SIMD (Single Instruction, Multiple Data Stream): One instruction applied to multiple data points simultaneously. Examples include graphics processing.

    • MISD (Multiple Instruction, Single Data Stream): Multiple instructions operate on the same data stream. Rare and used in specialized applications.

    • MIMD (Multiple Instruction, Multiple Data Stream): Multiple processors execute different instructions on different data streams concurrently. Widely used in multi-core and multiprocessor systems.

3. MIMD Analysis: SMP, Clusters, and NUMA

/

  • For each of SMP (Symmetric Multiprocessor), Clusters, and NUMA (Nonuniform Memory Access), address the following points:

    a. Organization/Architecture

    • Describe how the architecture is organized in terms of processors, memory, and data flow.

    b. Performance

    • Compare how each architecture performs, particularly under varying loads.

    c. Reliability

    • Discuss reliability in each system and any fault-tolerance or redundancy measures present.

    d. Scalability

    • Explain how each architecture scales with increasing numbers of processors or nodes.

    e. Potential Applications

    • Provide examples of real-world applications or industries that benefit from each architecture.

4. Conclusion

  • Summarize the key findings from your analysis of parallel architectures.

  • Conclude by discussing how each architecture is suited to different types of tasks and its relevance in current and future computing.

  • For writing an effective conclusion, consider including a brief reflection on the impact of parallel processing technology and its potential future developments.

Task Division:

  1. Introduction and Conclusion:

    • Assign Person 1 to handle both the Introduction and Conclusion sections.

    • They can provide a cohesive start and end to the essay, summarizing the group’s findings and ensuring the essay has a consistent tone.

  2. Taxonomy of Parallel Processor Architectures:

    • Assign Person 2 to research and write about SISD and SIMD.

    • Assign Person 3 to handle MISD and MIMD.

    • Each person can describe the definitions, characteristics, and examples of these architectures.

  3. MIMD Analysis: SMP, Clusters, and NUMA:

    • Assign Person 4 to cover SMP and Clusters (for all aspects like Organization/Architecture, Performance, Reliability, Scalability, and Potential Applications).

    • Assign Person 5 to analyze NUMA and ensure the format is consistent with Person 4’s analysis.

    • NOTE Critically analyse each of it for this step for HIGH marks

Suggested Work Flow:

  • After completing individual sections, Person 1 can review the full draft and tie everything together in the conclusion, while also checking for overall consistency.

  • If any sections are particularly challenging, group members can provide extra support to balance the workload.

Timeline and Deliverables

Week 1: Research and Initial Drafting

  • Goal: Complete research and start drafting each section.

  • Deliverables:

    • End of Week 1: Each person should submit their research notes and a rough outline of their assigned section to the group.

    • Due Date: By the end of the week.

  • Notes: Allocate time for group discussions or check-ins to share findings and address any questions.

Week 2: First Drafts of Each Section

  • Goal: Have each person submit a first draft of their section (Introduction, taxonomy parts, MIMD analysis sections, and Conclusion).

  • Deliverables:

    • Mid-Week 2: Each person submits their full draft to the group.

    • End of Week 2: Group meeting to review all drafts and ensure consistency in tone and formatting.

  • Notes: Assign one person to compile the drafts into a single document for easier revision later.

Week 3: Revisions and Additional Research

  • Goal: Revise drafts based on group feedback and add any needed research.

  • Deliverables:

    • Mid-Week 3: Revised drafts of each section, based on feedback from Week 2.

    • End of Week 3: Complete any additional research needed and finalize the core content.

  • Notes: This is the time to fine-tune content, confirm references, and verify that all instructions for each section are covered.

Week 4: Final Draft and Proofreading

  • Goal: Assemble the final draft, proofread, and make formatting adjustments.

  • Deliverables:

    • Mid-Week 4: Group proofreading session (ideally done by two people), ensuring all parts are consistent in style and flow.

    • End of Week 4: Submit the final version to all group members for final checks and approval.

  • Notes: Consider formatting, clarity, and organization. Everyone should approve the final version by the end of the week.

Final Week (Buffer Week): Final Edits and Submission

  • Goal: Have a polished essay ready for submission.

  • Deliverables:

    • Early in the Week: Make any minor adjustments or updates based on the final review.

    • Mid-Week: Confirm that all sections, citations, and formatting meet requirements.

    • End of the Week: Submit the assignment.