What is the primary difference between Multicore and Multiprocessor systems?

A multicore system integrates multiple cores onto a single physical chip (die), sharing the same socket and often a shared cache. A multiprocessor system consists of multiple separate physical chips, each in its own socket, connected via a system bus or network.

How does Data Parallelism differ from Task Parallelism?

Data Parallelism involves performing the same operation on different pieces of data simultaneously (e.g., adding two vectors). Task Parallelism involves performing different operations or functions simultaneously on the same or different data (e.g., one core handles UI while another handles background calculations).

Compare UMA (Uniform Memory Access) and NUMA (Non-Uniform Memory Access).

In UMA, all processors share physical memory with the same access time. In NUMA, memory access time depends on the memory location relative to the processor; a core accesses its local memory faster than memory attached to another processor.

What is a 'Race Condition' in parallel programming?

A race condition occurs when the output of a program depends on the uncontrolled timing or sequence of thread execution. It typically happens when multiple threads attempt to read and write to a shared variable simultaneously without proper synchronization.

What is 'False Sharing' and why does it degrade performance?

False sharing occurs when different cores modify different variables that happen to reside on the same cache line. The cache coherence protocol forces the cache line to be moved back and forth between cores, causing unnecessary overhead even though the data being modified is logically independent.

What is a 'Deadlock' in a multicore environment?

A deadlock is a state where two or more threads are unable to proceed because each is waiting for the other to release a resource. For example, Thread A holds Lock 1 and waits for Lock 2, while Thread B holds Lock 2 and waits for Lock 1.

Define a 'Core' in the context of computer architecture.

A core is an independent processing unit that contains its own ALU, control unit, and registers. It is capable of fetching, decoding, and executing instructions autonomously within a larger multicore processor.

What is 'Speedup' in parallel computing?

Speedup is a measure of the performance gain achieved by using a parallel system compared to a sequential one. It is calculated as the ratio of the execution time on a single processor to the execution time on $N$ processors.

What does the 'M' stand for in the MESI cache coherence protocol?

The 'M' stands for 'Modified.' It indicates that the cache line has been modified by the local core and is inconsistent with main memory; the local core now has the only valid copy and is responsible for writing it back.

Why does Amdahl's Law suggest that adding more cores has diminishing returns?

Because every program has a sequential portion that cannot be parallelized. As the number of cores increases, the time spent on the parallel portion shrinks, but the time spent on the sequential portion remains constant, eventually dominating the total execution time.

Multicore & Parallel Processors | OCR AS-Level Computer Science

1. The Characteristics of Contemporary Processors, Input, Output & Storage Devices

Multicore & Parallel Processors

Summary

Multicore and parallel processing represent a shift in computer architecture from increasing single-core clock speeds to integrating multiple processing units on a single chip. This approach leverages hardware-level parallelism to improve throughput and energy efficiency, though it introduces significant challenges in memory consistency, synchronization, and software scalability.

1. Definition & Core Concepts

Multicore Processors: A multicore processor is an integrated circuit to which two or more independent processing units, called cores, have been attached for enhanced performance and reduced power consumption. Each core can independently execute program instructions, allowing the system to perform multiple tasks simultaneously or accelerate a single task by dividing it into parallel sub-tasks.
Parallelism vs. Concurrency: While often used interchangeably, parallelism refers to the simultaneous execution of multiple tasks on different hardware resources, whereas concurrency is the management of multiple tasks that start, run, and complete in overlapping time periods. In a multicore system, true parallelism is achieved when different cores execute different threads at the exact same moment.
Hardware Threads: Many modern cores support Simultaneous Multithreading (SMT), which allows a single physical core to appear as multiple logical processors to the operating system. This technique improves core utilization by executing instructions from a second thread while the first thread is stalled, such as during a memory access.

Diagram of a typical multicore architecture showing individual cores with private L1/L2 caches connected via an interconnect bus to a shared L3 cache and main memory.

2. Underlying Principles: Flynn's Taxonomy

SISD (Single Instruction, Single Data): This is the traditional sequential computer where a single processor executes one instruction stream to operate on a single data stream. It represents the baseline for performance comparison before parallel enhancements are applied.
SIMD (Single Instruction, Multiple Data): In this model, a single instruction is broadcast to multiple processing elements, each operating on different data points simultaneously. This is highly effective for data-parallel tasks like image processing or scientific simulations where the same operation is applied to a large array.
MIMD (Multiple Instruction, Multiple Data): This is the most common model for multicore processors, where each core can execute different instructions on different data. It provides the greatest flexibility, allowing the system to run entirely different programs or different parts of a single complex program in parallel.

3. Performance Scaling: Amdahl's Law

4. Memory Hierarchy & Cache Coherence

5. Key Distinctions

6. Exam Strategy & Tips