Optimization of Binomial Option Pricing on Intel MIC Heterogeneous System
In these years, computerization has been more and more important in the financial area. The computational intensity and real-time constraints of those financial models require high-throughput parallel architectures. In this paper, optimization of widely-used binomial option pricing model has been implemented on the worlds largest supercomputer, Tianhe-2. In our work, we employ several optimizing techniques to efficiently utilize the architecture of Intel MIC heterogeneous system to improve the performance. The experimental results show that, compared with the serial implementation, the optimized binomial option pricing achieves 33X speedup on one Intel Xeon CPU and 61X speedup on one Intel Xeon Phi coprocessor. Further experiments on Intel MIC heterogeneous system indicate that our implementation attains a speed-up factor of 254 on one Tianhe-2 computing node.
KeywordsBinomial option pricing MIC Parallel process Heterogeneous system Optimization
We thank the anonymous reviewers for their valuable comments. This work is supported financially by the National Hi-tech Research and Development Program of China under contracts 2012AA010902.
- 1.Intel MIC Architecture. https://software.intel.com/en-us/articles/intel-xeon-phi-coprocessor-codename-knights-corner
- 2.The message passing interface (mpi) standard. http://www.mcs.anl.gov/research/projects/mpi/
- 3.The openmp api specification for parallel programming. http://openmp.org
- 4.Corden, M.J., Kreitzer, D.: Consistency of floating-point results using the intel compiler or why doesnt my application always give the same answer. Technical report, Intel Corporation, Software Solutions Group (2009)Google Scholar
- 6.Dixon, M., Chong, J., Keutzer, K.: Acceleration of market value-at-risk estimation. In: Proceedings of the 2nd Workshop on High Performance Computational Finance, p. 5. ACM (2009)Google Scholar
- 7.Dongarra, J.: Visit to the National University for Defense Technology Changsha. University of Tennessee 199, China (2013). http://www.netlib.org/utk/people/JackDongarra/PAPERS/tianhe-2-dongarra-report.pdf Google Scholar
- 11.Jeffers, J., Reinders, J.: Intel Xeon Phi Coprocessor High-Performance Programming. Newnes, Boston (2013)Google Scholar
- 12.Jin, Q., Thomas, D.B., Luk, W., Cope, B.: Exploring reconfigurable architectures for tree-based option pricing models. ACM Trans. Reconfigurable Technol. Syst. (TRETS) 2(4), 2586–2600 (2009). Article No. 21Google Scholar
- 14.Malladi, R.K.: Using intel vtune performance analyzer events/ratios optimizing applications (2009)Google Scholar
- 15.Newburn, C.J., Dmitriev, S., Narayanaswamy, R., Wiegert, J., Murty, R., Chinchilla, F., Deodhar, R., McGuire, R.: Offload compiler runtime for the intel xeon phi coprocessor. In: Parallel and Distributed Processing Symposium Workshops & Ph.D. Forum (IPDPSW), 2013 IEEE 27th International, pp. 1213–1225. IEEE (2013)Google Scholar
- 17.Thulasiram, R.K., Dondarenko, D.: Performance evaluation of parallel algorithms for pricing multidimensional financial derivatives. In: International Conference on Parallel Processing Workshops, 2002, Proceedings, pp. 306–313. IEEE (2002)Google Scholar