Network Architecture - Unveiled

Network Architecture - Unveiled

SMP versus MSMP

Memory Access

Master-Slave Multi Processing lies in the category of Asymmetric Multiprocessing systems. In MSMP systems, the master has the right to access the memory and distribute it to slave processors. One of the drawbacks with this configuration is that most of the times the master acts like a bottleneck which cause delay in task executions. On the other hand SMP systems, each of the processors have distribution and access to the memory. However, this creates the problem of resources conflict especially the memory.

I/O Resource Access

As far as MSMP systems are under consideration, the master processor takes care of various I/O resource operations as well as the access of various systems. However, on the other hand, if a master processor is busy, the slave processors gain the liberty to get the task done as per their requirements. On the contrary, the SMP has equal resource access to various input as well as output resources. Hence, the same problem of resource conflicts exists here too.

Instruction Execution Throughput

For the master-slave arrangement of processors, the parallel execution for a single task is conducted by the help of located subtasks to multiple processes simultaneously. As the master is only responsible for executing the task, this makes the system quite simple and efficient to operate. On the other hand, the Symmetric Multi Processing systems have the conflict on resource allocation. Hence, a method called 'floating meter' is used to make a processor to take the lead and execute the task. Once that is done, the operating system floats from one processor to another causing parallel execution of various applications a success. Processor allocation makes it possible to do the assigning to the processors which are idle and this process goes on until all the tasks are executed, or the processors are busy to handle any other information (Russinovich and Solomon et al., 2005 ; Anvari.net, 2003).

Hyperthreading or Not

Hyperthreading technology is the gateway that allows users to use resources possessed by their respective processors and which are under idle condition or out of use due to the current thread execution. In some cases, it is coherent to the symmetric multiprocessing (SMP). Why is it, so? It is because, in multiprocessors, we have the virtual processor coexisting with the bus of real (more specifically physical) multiprocessors. As an example, consider a dual core processing system having the hyperthreaded processors. Hence, most of the time it appears to be four processors as far as programs are under consideration. Using Hyperthreading a dual core processor appears to be a processor having four symmetric processors. As far as most of the high-performance computing systems are concerned, Hyperthreading is sometimes useful while sometimes it is not. In SMP systems, the operating system has the ability to process two or more tasks simultaneously. Having the two CPUs reduces the time of execution of various tasks and users do not have to wait for much longer times. At the throughput, we get a system which ...