Abstract | ||
---|---|---|
Hydra is a chip multiprocessor (CMP) with integrated support for thread-level speculation. Thread-level speculation provides a way to parallelize sequential programs without the need for data depen- dence analysis or synchronization. This makes it possible to paral- lelize applications for which static memory dependence analysis is difficult or impossible. While performance of the baseline Hydra system on applications with medium to large grain parallelism is good, the performance on integer applications with fine-grained parallelism is unimpressive. In this paper, we describe a collection of software and hardware techniques for improving speculation per- formance over the baseline speculative Hydra CMP. These tech- niques focus on reducing the overheads associated with speculation and improving the speculation behavior of the applications using code restructuring. When these techniques are applied to a set of eleven integer, multimedia and floating-point benchmarks, signifi- cant performance improvements result. In particular, the overall performance of the integer benchmarks is improved by seventy-five percent. |
Year | DOI | Venue |
---|---|---|
1999 | 10.1145/2591635.2667177 | International Conference on Supercomputing 2006 |
Keywords | DocType | ISBN |
hydra cmp,chip multiprocessor,design,parallelism and concurrency,parallel programming,feedback-driven optimization,speculatively parallel application,measurement,data speculation,theory,multithreading,performance evaluation,threads,performance,data processing,sequences,floating point,thread level speculation,dependence analysis | Conference | 1-58113-164-X |
Citations | PageRank | References |
38 | 4.16 | 10 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Kunle Olukotun | 1 | 4532 | 373.50 |
Lance Hammond | 2 | 520 | 66.61 |
Mark Willey | 3 | 38 | 4.49 |