The Paradigm Shift to Multi-Cores: Opportunities and Challenges
Per Stenström · 2007
The field of computer architecture is a truly amazing one. On the one hand, technologists offer an increase in performance annually, on the other hand, software designers set up rules for how we can harness the ever increasing computational performance improvements. Right now, we are in the beginning of an interesting development in which we can no longer rely on making a single processor faster; parallel processing is likely our only hope to continue delivering a doubling of processor performance every eighteen month. To this end, this article discusses the opportunities and challenges of the new breed of multi-core computers that is one of the important focal points of the developments in computer architecture over the next years to come. Stamatis Vassiliadis, who offered his leadership to launch the EU funded SARC project, wisely set the direction of it to explore scalability issues of multi-cores. As one of his work package leaders, I will focus on issues related to scalability of the memory system and how ease of programming puts specific demands on the semantics offered by future multi-core memory systems. 1.