ABSTRACT
The availability of highly parallel computing platforms based on multi and manycore processors enables a wide variety of technical solutions for systems across the embedded and high-performance computing domains. Different application domains can be characterised by their typical levels of dynamicity. This chapter considers allocation mechanisms at both grid and manycore CPU level, and reviews the most significant trends and achievements in terms of guaranteed performance and energy efficiency. Most allocation techniques addressing energy efficiency operate at the many-core processor level, mainly because of the difficulties of dealing with energy-related metrics at larger system granularities. While the approaches have presented sophisticated resource allocation approaches that can provide performance guarantees and/or improve energy efficiency, there are still challenges that require more advanced resource allocation approaches. The chapter describes some of those challenges.
