Boxun Li
ABSTRACT
The invention of resistive-switching random access memory (RRAM) devices and RRAM crossbar-based computing system (RCS) demonstrate a promising solution for better performance and power efficiency. The interfaces between analog and digital units, especially AD/DAs, take up most of the area and power consumption of RCS and are always the bottleneck of mixed-signal computing systems. In this work, we propose a novel architecture, MEI, to minimize the overhead of AD/DA by MErging the Interface into the RRAM crossbar. An optional ensemble method, the Serial Array Adaptive Boosting (SAAB), is also introduced to take advantage of the area and power saved by MEI and boost the accuracy and robustness of RCS. On top of these two methods, a design space exploration is proposed to achieve trade-offs among accuracy, area, and power consumption. Experimental results on 6 diverse benchmarks demonstrate that, compared with the traditional architecture with AD/DAs, MEI is able to save 54.63%~86.14% area and reduce 61.82%~86.80% power consumption under quality guarantees; and SAAB can further improve the accuracy by 5.76% on average and ensure the system performance under noisy conditions.
- H. Esmaeilzadeh, A. Sampson, L. Ceze, and D. Burger, "Neural acceleration for general-purpose approximate programs," in MICRO, 2012, pp. 449--460. Google Scholar
Digital Library
- S. A. McKee, "Reflections on the memory wall," in Proceedings of the 1st conference on Computing frontiers. ACM, 2004. Google ScholarDigital Library
- B. Liu et al., "Reduction and ir-drop compensations techniques for reliable neuromorphic computing systems," in ICCAD, 2014. Google ScholarDigital Library
- C. Xu, X. Dong, N. P. Jouppi, and Y. Xie, "Design implications of memristor-based rram cross-point structures," in DATE, 2011.Google Scholar
- M. Versace and B. Chandler, "Moneta: a mind made from memristors," IEEE Spectrum, vol. 47, no. 12, pp. 30--37, 2010. Google ScholarDigital Library
- B. Li, Y. Shan et al., "Memristor-based approximated computation," in ISLPED, 2013, pp. 242--247. Google ScholarDigital Library
- R. St. Amant, A. Yazdanbakhsh et al., "General-purpose code acceleration with limited-precision analog computation," in ISCA, 2014, pp. 505--516. Google ScholarDigital Library
- X. Liu, M. Mao et al., "A heterogeneous computing system with memristor-based neuromorphic accelerators," in IEEE High Performance Extreme Computing (HPEC), 2014.Google Scholar
- S. Yu, B. Gao et al., "A low energy oxide-based electronic synaptic device for neuromorphic visual systems with tolerance to device variation," Advanced Materials, 2013.Google Scholar
- M. Hu, H. Li et al., "Hardware realization of bsb recall function using memristor crossbar arrays," in DAC, 2012, pp. 498--503. Google ScholarDigital Library
- B. Li, Y. Wang et al., "Ice: inline calibration for memristor crossbar-based computing engine," in DATE, 2014, pp. 184--187. Google ScholarDigital Library
- Y. Deng, H.-Y. Chen et al., "Design and optimization methodology for 3d rram arrays," in IEEE International Electron Devices Meeting (IEDM), 2013, pp. 9--11.Google Scholar
- W.-H. Tseng and P.-C. Chiu, "A 960ms/s dac with 80db sfdr in 20nm cmos for multi-mode baseband wireless transmitter," in VLSI Circuits Digest of Technical Papers, 2014 Symposium on. IEEE, 2014, pp. 1--2.Google Scholar
- J. Proesel, G. Keskin et al., "An 8-bit 1.5 gs/s flash adc using post-manufacturing statistical selection," in CICC, 2010.Google Scholar
- C. M. Bishop et al., "Neural networks for pattern recognition," 1995. Google ScholarCross Ref
- Z.-H. Zhou, Ensemble methods: foundations and algorithms. CRC Press, 2012. Google ScholarCross Ref
- ITRS, "International technology roadmap for semiconductors," 2013.Google Scholar
Index Terms
- Merging the interface: power, area and accuracy co-optimization for RRAM crossbar-based mixed-signal computing system
Recommendations
Power-driven flip-flop merging and relocation
ISPD '11: Proceedings of the 2011 international symposium on Physical designWe propose a power-driven flip-flop merging and relocation approach that can be applied after conventional timing-driven placement and before clock network synthesis. It targets to reduce the clock network size and thus the clock power consumption, as ...
Slack budgeting and slack to length converting for multi-bit flip-flop merging
DATE '13: Proceedings of the Conference on Design, Automation and Test in EuropeIn this paper we propose a flexible slack budgeting approach for post-placement multi-bit flip-flop (MBFF) merging. Our approach considers existing wiring topology and flip-flop delay changes for achieving more accurate slack budgeting. Besides, we ...
Power Aware Interface Synthesis for Bus-Based SoC Designs
DATE '04: Proceedings of the conference on Design, automation and test in Europe - Volume 2In this paper we discuss the problem of interface synthesis for a system on a chip (SoC) such that the power consumption is minimized under some given latency constraints. Since the AMBA protocol has become one of the standard interfaces for SoC cores, ...
Comments