Large-Scale Shared-Memory Multiprocessors

 

Bus inevitably becomes a bottleneck when many processors are used

• Use a more general interconnection network, so snooping does not work
• DRAM memory is also distributed
• Each node allocates space from local DRAM
• Copies of remote data are made in cache

Major design issues:

• How to find and represent the ``directory'' of each line?
• How to find a copy of a line?

As a case study, we will look at S3.MP (Sun's Scalable Shared memory Multi-Processor, a CC-NUMA (cache-coherent non-uniform memory access) architecture

• Case Study: S3.MP
• Protocol Basics
• Read Requests
• ...
• Writes
• Replacements
• Summary and Conclusions