Dynamic Thermal Management for Heterogeneous Multi-Core Processor Throughput Optimization

Three-dimensional heterogeneous multi-core processor (3D MCP) has become an emerging technology because of advantages of higher packaging density and more flexible in heterogeneous integrations. Nevertheless, due to the inherent heat removal limitation in 3D ICs, thermal issues must be considered when maximizing the throughput of 3D MCPs. Furthermore, since the thermal behavior of a core in 3D MCPs strongly depends on its location, a proper task allocation can alleviate potential thermal problems and improve the performance. However, conventional techniques require computationally intensive thermal simulation, which prohibits its usage from the on-line application. In this paper, an efficient dynamic thermal management technique is proposed to solve the problem of throughput optimization under thermal constraints for 3D MCPs. Our proposed method achieves a 20.82X runtime speedup shown in experimental results, which are comparable to the exhaustive solutions obtained from optimization-modeling software LINGO. Moreover, our throughput results, with and without consideration of unfinished tasks, are only 4.39% and 0.69% worse, respectively than that of the exhaustive method. Note that, our method takes only 0.951 ms, which is 59.39 times faster than that of the previous work in 128 task-to-core allocations.

並列關鍵字

Dynamic thermal management ； heterogeneous multi-core processor ； migration penalty ； task allocation ； task migration ； thermal aware ； three-dimensional integration ； throughput optimization ； temperature uniformity

國際替代計量

全文下載

主題瀏覽