Question

The processors in a distributed-memory system, receives data from the memory of other processors by requesting the processor to send a copy of the data. In a shared memory system, a processor can read from any of the k memory modules by connecting to the memory using the interconnection network.

Which interconnection network is best for a distributed memory system? Justify your answer; you can do this by comparing three parameters of the remaining 7 interconnection networks with your selected network.

Answer 1

The best interconnection network for a distributed memory system would be one that supports high bandwidth and low latency.

Parameters to compare interconnection networks include bandwidth, latency, and scalability.

Step-by-step explanation:

In a distributed memory system, one such network is the hierarchical bus network, where the processors are connected to a common bus.

This allows for direct communication between processors, reducing latency.

Another option is the crossbar network, which provides full connectivity between all processors, enabling efficient data transfer.

Three parameters to consider when comparing the interconnection networks are:

1. Bandwidth: How much data can be transferred per unit of time. The hierarchical bus network may have lower bandwidth compared to the crossbar network.
2. Latency: The time taken for a message to travel from the source to the destination. The hierarchical bus network may have higher latency compared to the crossbar network.
3. Scalability: The ability of the network to handle a large number of processors. The crossbar network may be more scalable compared to the hierarchical bus network.