Dom. Mod. Final Exam-Parallel Distributing

What is the primary reason parallel computing offers a solution to the ever-increasing demand for computational power in today's world?

• It simplifies the design of sequential algorithms.

• It eliminates the need for large memory systems.

• It allows multiple tasks or parts of a single task to be executed simultaneously.

• It reduces the cost of traditional sequential computers.

Which computer architecture classification, proposed by Michael J. Flynn, is characterized by a single control unit fetching a single instruction stream and operating on a single data stream?

• SIMD (Single Instruction, Multiple Data)

• MISD (Multiple Instruction, Single Data)

• MIMD (Multiple Instruction, Multiple Data)

• SISD (Single Instruction, Single Data)

Which classification in Flynn's Taxonomy applies a single instruction to multiple data streams concurrently and is common in vector processors and GPUs?

• SISD

• SIMD

• MISD

• MIMD

The most popular model for general-purpose parallel computers, enabling high flexibility and scalability by having multiple processors execute different instructions on different data concurrently, is known as:

• SIMD

• MISD

• MIMD

• SISD

Which of Flynn's classifications is less common in general-purpose computing but can be found in specialized systems like systolic arrays, where multiple functional units perform different operations on the same data stream?

• MIMD

• MISD

• SIMD

• SISD

In a Shared Memory Architecture, how does communication between processors occur?

• Explicitly via message passing.

• Through a high-speed network switch.

• Implicitly via a single, global memory address space.

• Only through I/O channels.

Which memory architecture is known for being highly scalable and avoids memory contention issues because each processor has its own private, local memory?

• Shared Memory Architectures

• Distributed Memory Architectures

• Hybrid Memory Architectures

• Uniform Memory Access (UMA)

What is the main disadvantage of Shared Memory Architectures?

• Scalability issues due to memory contention and cache coherence problems.

• Programs require complex explicit data distribution.

• It limits the number of instruction streams.

• Communication requires explicit message passing.

Communication in Distributed Memory Architectures requires:

• Implicit access to a unified address space.

• Simultaneous memory access handling.

• Explicit message passing.

• Uniform memory access.

Architectures that combine aspects of both shared and distributed memory (e.g., clusters where each node is a shared-memory multiprocessor) are known as:

• Massively Parallel Processors (MPP)

• Symmetric Multiprocessors (SMP)

• Hybrid Memory Architectures

• PRAM Models