Slides-Cap8-Part1-Parallel Architectures Quiz

<p>Symmetric multiprocessors</p> Signup and view all the answers

<p>Multiple instruction streams, multiple data streams</p> Signup and view all the answers

<p>Nothing</p> Signup and view all the answers

<p>Increasing memory bandwidth</p> Signup and view all the answers

<p>Processors share a single memory with uniform access times</p> Signup and view all the answers

<p>Complex communication among processors</p> Signup and view all the answers

<p>SMP has uniform memory access, while DSM has non-uniform memory access</p> Signup and view all the answers

<p>Reduced hardware requirements</p> Signup and view all the answers

<p>Providing locality and replication</p> Signup and view all the answers

<p>Processors with independent memories and address spaces</p> Signup and view all the answers

<p>Implicit communication through memory access</p> Signup and view all the answers

<p>Bigger market share, both in dollars and units</p> Signup and view all the answers

<p>Clusters for systems on the internet</p> Signup and view all the answers

<p>Tightly coupled processors, controlled by a single OS, sharing memory through a shared address space</p> Signup and view all the answers

<p>Options include sharing cache, memory, and I/O system, or memory and I/O system, or I/O system, or nothing, typically communicating through networks</p> Signup and view all the answers

<p>Symmetric (shared-memory) multiprocessors - SMP and distributed shared memory - DSM</p> Signup and view all the answers

<p>Up to 32 cores, single centralized memory, uniform access time, and potential bottleneck in memory/bus</p> Signup and view all the answers

<p>Avoiding bottlenecks in memory/bus</p> Signup and view all the answers

<p>Equal access to all of the memory from all of the processors</p> Signup and view all the answers

<p>DSM increases bandwidth and reduces access latency by distributing memory to multiple nodes, allowing each node to be a small distributed system with centralized memory.</p> Signup and view all the answers

<p>SMP has uniform access times to memory, while DSM has non-uniform memory access (NUMA) times, depending on the location of the data word in memory.</p> Signup and view all the answers

<p>Multicomputers have independent memories and address spaces, and communicate through interconnection networks.</p> Signup and view all the answers

<p>The message passing model allows for explicit communication between processors, which is necessary for synchronization in distributed memory systems.</p> Signup and view all the answers

<p>Caches provide locality and replication, reducing the need for memory bandwidth and increasing the efficiency of SMP architectures.</p> Signup and view all the answers

<p>SMP architectures require less extra hardware and are based on general-purpose processors, making them more cost-effective.</p> Signup and view all the answers

<p>SMP architectures have a larger market share in terms of both dollars and units, while multicomputers are popular in cluster systems for internet-based systems.</p> Signup and view all the answers

<p>The primary challenge is the increased effort required in software to take advantage of the increased memory bandwidth.</p> Signup and view all the answers

<p>The implicit communication model used in SMP architectures simplifies synchronization, while the explicit communication model used in DSM architectures introduces additional synchronization challenges.</p> Signup and view all the answers

<p>OpenMP provides a thread-based model for parallel programming, allowing for efficient use of multiple processors in SMP architectures.</p> Signup and view all the answers