The Community for Technology Leaders
2017 IEEE International Symposium on Workload Characterization (IISWC) (2017)
Seattle, WA, USA
Oct. 1, 2017 to Oct. 3, 2017
ISBN: 978-1-5386-1234-7
pp: 135-145
Hiroshi Sasaki , Department of Computer Science, Columbia University
Fang-Hsiang Su , Department of Computer Science, Columbia University
Teruo Tanimoto , Graduate School of Information Science and Electrical Engineering, Kyushu University
Simha Sethumadhavan , Department of Computer Science, Columbia University
ABSTRACT
Designing and optimizing computer systems require deep understanding of the underlying system. Historically many important observations that led to the development of essential hardware and software optimizations were driven by empirical studies of program behavior. In this paper we report an interesting property of dynamic program execution by viewing it as a changing (or social) network. In a program social network, two instructions are friends if there is a producer-consumer relationship between them. One prominent result is that the outdegree of instructions follow heavy tails or power law distributions, i.e., a few instructions produce values for many instructions while most instructions do so for very few instructions. In other words, the number of instruction dependencies is highly skewed. In this paper we investigate this curious phenomenon. By analyzing a large set of workloads under different compilers, compilation options, ISAs and inputs we find that the dependence skew is widespread, suggesting that it is fundamental. We also observe that the skew is fractal across time and space. Finally, we describe conditions under which skew emerges within programs and provide evidence that suggests that the heavy-tailed distributions are a unique program property.
INDEX TERMS
Registers, Benchmark testing, Optimization, Social network services, C++ languages, Libraries, Probability distribution
CITATION

H. Sasaki, F. Su, T. Tanimoto and S. Sethumadhavan, "Why do programs have heavy tails?," 2017 IEEE International Symposium on Workload Characterization (IISWC), Seattle, WA, USA, 2017, pp. 135-145.
doi:10.1109/IISWC.2017.8167771
209 ms
(Ver 3.3 (11022016))