Algorithms and Parallel Computing

4.1 INTRODUCTION

Shared-memory processors are popular due to their simple and general programming model, which allows simple development of parallel software that supports sharing of code and data [27].

Shared-memory processors provide a single physical address space for all processors, and each processor can run its own program using its local memory and cache. The processors also have access to shared memory arranged in separate modules. The processors communicate using shared variables, which are stored in the shared memory to be accessible to any processor. Memory in a shared-memory multiprocessor system is organized in a hierarchical fashion as shown in Fig. 4.1. The figure shows a system with n processors and b shared memory modules. Each processor has its own internal registers, cache, and local memory. The cache stores all data currently used by the processor. The local memory stores local variables not meant to be shared with the other processors. The shared memory stores the shared variables that need to be exchanged between the processors. The interconnection network allows more than one processor to simultaneously access different shared memory modules through the network. This allows more than one memory read/write operations to take place simultaneously.

Figure 4.1 Shared-memory processors. Each processor has its local cache and local memory. All processors can access the shared memory modules through the interconnection network.

We explained in Chapter 2 that ...