Computer Organization and Design RISC-V Edition

Figure 6.13 Classic organization of a multiprocessor with multiple private address spaces, traditionally called a message-passing multiprocessor.Note that unlike the SMP in Figure 6.7, the interconnection network is not between the caches and memory but is instead between processor-memory nodes.

There have been several attempts to build large-scale computers based on high-performance message-passing networks, and they do offer better absolute communication performance than clusters built using local area networks. Indeed, many supercomputers today use custom networks. The problem is that they are much more expensive than local area networks ...