Instruction set extensions for dynamic time warping
Article No.: 18, Pages 1 - 10
Abstract
Processor specialization through application-specific instruction set customization can significantly improve performance while reducing energy. Due to the costs associated with semiconductor fabrication, specialized processors are only viable for products with high production volumes. The emergence of low-cost sensor-based computing products in recent years has created an urgent need to process time-series data with the utmost efficiency. Although most sensor data is fixed-point, the normalization process---an absolute necessity for highly accurate similarity search of time-series data---converts the data to floating-point in order to avoid a loss in precision. The sensors that collect time-series data are typically connected to low-power microcontrollers or RISC processors sans floating point units. The computational requirements of real-time similarity search would overwhelm such processors. To address this concern, we introduce a specialized instruction set for time-series data mining applications to a 32-bit embedded processor, yielding a 4.87x performance improvement and a 78% reduction in energy consumption compared to a highly optimized software implementation.
References
[1]
Ball, T., and Larus, J. R., "Optimally profiling and tracing programs," ACM Trans. Programming Languages and Systems, vol. 16, no. 4, July, 1994, pp. 1319--1360. DOI= http://dx.doi.org/10.1145/183432.183527
[2]
Chong, Y. J. and Parameswaran, S., "Custom floating-point unit generation for embedded systems," IEEE Trans. Computer-Aided Design of Integrated Circuits and Systems, vol. 28, no. 5, May, 2009, pp. 638--650. DOI= http://dx.doi.org/10.1109/TCAD.2009.2013999
[3]
Dally, W. J., et al., "Efficient embedded computing," Computer, vol. 41, no. 7, July 2008, pp. 27--32. DOI= http://dx.doi.org/10.1109/MC.2008.224
[4]
de Dinechin, F., and Pasca, B., "Designing custom arithmetic data paths with FloPoCo," IEEE Design and Test of Computers, vol. 28, no. 4, July-Aug. 2011, pp. 18--27. DOI= http://dx.doi.org/10.1109/MDT.2011.44
[5]
Ding, H., et al., "Querying and mining of time series data: experimental comparison of representations and distance measures," Proceedings of the VLDB Endowment, vol. 1, no. 2, Aug. 2008, pp. 1542--1552. URL= http://dl.acm.org/citation.cfm?id=1454226
[6]
Evans, D., "The Internet of Things: How the Next Evolution of The Internet Is Changing Everything," Cisco Internet Business Solutions Group. White Paper. April, 2011. URL= http://www.cisco.com/web/about/ac79/docs/innov/IoT_IBSG_0411FINAL.pdf
[7]
Fu, A. W-C., et al., "Scaling and time warping in time series querying," International Journal on Very Large Data Bases, vol. 17, no. 4, July, 2008, pp. 899--921. DOI= http://dx.doi.org/10.1007/s00778-006-0040-z
[8]
Hameed, R., et al., "Understanding sources of inefficiency in general-purpose chips," Comm. ACM, vol. 54, no. 10, Oct. 2011, pp. 85--93. DOI= http://dx.doi.org/10.1145/2001269.2001291
[9]
Hockert, N., and Compton, K., "Improving floating-point performance in less area: fractured floating point units (FFPUs)," Journal of Signal Processing Systems, vol. 67, no. 1, April, 2012, pp. 31--46. DOI= http://dx.doi.org/10.1007/s11265-010-0561-y
[10]
Iwata, T., et al., "A speech recognition processor," IEEE Int. Solid-State Circuits Conf. (ISSCC '83), pp. 120--121, Feb. 23--25, 1983, DOI= http://dx.doi.org/10.1109/ISSCC.1983. 1156535
[11]
Keogh, E., and Kasetty, S., "On the need for time series data mining benchmarks: a survey and empirical demonstration," Journal of Data Mining and Knowledge Discovery, vol. 7, no. 4, Oct. 2003, pp. 349--371. DOI= http://dx.doi.org/10.1023/A:1024988512476
[12]
Keogh, E., et al., "Supporting exact indexing of arbitrarily rotated shapes and periodic time series under Euclidean and warping distance measures," International Journal on Very Large Data Bases, vol. 18, no. 3, June, 2009, pp. 611--630. DOI= http://dx.doi.org/10.1007/s00778-008-0111-4
[13]
E. Keogh, et al., "The UCR Time Series Classification/Clustering Homepage," URL= http://www.cs.ucr.edu/~eamonn/time_series_data/
[14]
Kim, S-W., Park, S., and Chu, W. W., "An index-based approach for similarity search supporting time warping in large sequence databases," 17th Int. Conf. Data Engr. (ICDE '01), pp. 607--614, Apr. 2--6, 2001, DOI= http://dx.doi.org/10.1109/ICDE.2001.914875
[15]
Langhammer, M., "Floating-point datapath synthesis for FPGAs," Int. Conf. Field Programmable Logic and Applications (FPL '08), pp. 355--360, Sep. 8--10, 2008, DOI= http://dx.doi.org/10.1109/FPL.2008.4629963
[16]
Lemire, D., "Faster retrieval with a two-pass dynamic-time-warping lower bound," Pattern Recognition, vol. 42, no. 9, Sep. 2009, pp. 2169--2180. DOI= http://dx.doi.org/10.1016/j.patcog.2008.11.030
[17]
Loftian, R., and Jafari, R., "An ultra-low power hardware accelerator architecture for wearable computers using dynamic time warping," Design Automation and Test in Europe (DATE '13), pp. 913--916, Mar. 18--22, 2013, DOI= http://dx.doi.org/10.7873/DATE.2013.192
[18]
Loftian, R., and Jafari, R., "A low power wake-up circuitry based on dynamic time warping for body sensor networks," Int. Conf. Body Sensor Networks (BSN '11), pp. 83--88, May 23--25, 2011, DOI= http://dx.doi.org/10.1109/BSN.2011.43
[19]
Lowy, M., et al., "An architecture for a speech recognition system," IEEE Int. Solid-State Circuits Conf. (ISSCC '83), pp. 118--119, Feb. 23--25, 1983, DOI= http://dx.doi.org/10.1109/ISSCC.1983.1156528
[20]
Owen, R. E., "A VLSI dynamic time warp processor for connected and isolated word speech recognition," IEEE Int. Conf. Acoustics, Speech, and Signal Processing (ICASSP '85), pp. 985--988, Apr. 26--29, 1985, DOI= http://dx.doi.org/10.1109/ICASSP.1985.1168159
[21]
Pozzi, L., Atasu, K., and Ienne, P., "Exact and approximate algorithms for the extension of embedded processor instruction sets," IEEE Trans. CAD of Integrated Circuits and Systems, vol. 25, no. 7, July 2006, pp. 1209--1229. DOI= http://dx.doi.org/10.1109/TCAD.2005.855950
[22]
Quenot, G., et al., "A dynamic time warp VLSI processor for continuous speech recognition," IEEE Int. Conf. Acoustics, Speech, and Signal Processing (ICASSP '86), pp. 1549--1552, Apr. 7--11, 1988, DOI= http://dx.doi.org/10.1109/ICASSP.1986.1168945
[23]
Rakthanmanon, T., et al., "Searching and mining trillions of time series subsequences under dynamic time warping," 18th ACM SIGKDD Int. Conf. Knowledge Discovery and Data Mining (KDD '12), pp. 262--270, Aug. 12--16, 2012 DOI= http://dx.doi.org/10.1145/2339530.2339576
[24]
Sakoe, H., and Chiba, S. "Dynamic programming optimization for spoken word recognition," IEEE Trans. Acoustics, Speech, and Signal Processing, vol. 26, no. 1, Feb. 1978, pp. 43--49, DOI= http://dx.doi.org/10.1109/TASSP.1978.1163055
[25]
Sart, D., et al., "Accelerating dynamic time warping subsequence search with GPUs and FPGAs," 10th IEEE Int. Conf. Data Mining (ICDM '10), pp. 1001--1006, Dec. 13--17, 2010, DOI= http://dx.doi.org/10.1109/ICDM.2010.21
[26]
Sethia, A., et al., "A customized processor for energy efficient scientific computing," IEEE Trans. Computers, vol. 61, no. 12, Dec. 2012, pp. 1711--1723. DOI= http://dx.doi.org/10.1109/TC.2012.144
[27]
Wang, Z., et al., "Accelerating subsequence similarity search based on dynamic time warping distance with FPGA," ACM/SIGDA Int. Symp. Field Programmable Gate Arrays (FPGA '13), pp. 53--62, Feb. 11--13, 2013, DOI= http://dx.doi.org/10.1145/2435264.2435277
[28]
Xilinx, "Virtex-6 User Guide Xilinx Corporation," 2010.
[29]
Xilinx, "MicroBlaze Processor Reference Guide," October 2010.
[30]
Xilinx, "LogiCORE IP Fast Simplex Link (FSL) V20 Bus (v2.11c)," April 2010.
[31]
Xilinx, "Xilinx Power Tools Tutorial UG733 (v14.2)," July 25 2012.
Index Terms
- Instruction set extensions for dynamic time warping
Recommendations
Extended Instruction Exploration for Multiple-Issue Architectures
Regular PapersIn order to satisfy the growing demand for high-performance computing in modern embedded devices, several architectural and microarchitectural enhancements have been implemented in processor architectures. Extended instruction (EI) is often used for ...
Selection of instruction set extensions for an FPGA embedded processor core
IPDPS'06: Proceedings of the 20th international conference on Parallel and distributed processingA design process is presented for the selection of a set of instruction set extensions for the PowerPC 405 processor that is embedded into the Xilinx Virtex Family of FPGAs. The instruction set of the PowerPC 405 is extended by selecting additional ...
Automatic custom instruction identification for application-specific instruction set processors
The application-specific instruction set processors (ASIPs) have received more and more attention in recent years. ASIPs make trade-offs between flexibility and performance by extending the base instruction set of a general-purpose processor with custom ...
Comments
Information & Contributors
Information
Published In
Sponsors
Publisher
IEEE Press
Publication History
Published: 29 September 2013
Check for updates
Author Tags
Qualifiers
- Research-article
Conference
Acceptance Rates
CODES+ISSS '13 Paper Acceptance Rate 31 of 111 submissions, 28%;
Overall Acceptance Rate 280 of 864 submissions, 32%
Upcoming Conference
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 127Total Downloads
- Downloads (Last 12 months)0
- Downloads (Last 6 weeks)0
Reflects downloads up to 10 Aug 2024
Other Metrics
Citations
View Options
Get Access
Login options
Check if you have access through your login credentials or your institution to get full access on this article.
Sign in