ABSTRACT
A compact and low energy circuitry of time-based stochastic computing (TBSC) have been designed. In the TBSC theory, stochastic numbers (SNs) are represented by duty-cycle of periodic signals. Additionally, multiplication and addition operations of the SNs require the signals to be in-harmonic and uncorrelated in frequency. In order to dynamically tune the frequency, a current-starved structure is applied in a three-stage inverter chain type of oscillator with the neuron-MOS mechanism. By feeding the pulses to a neuron-MOS based inverter with adjustable switching threshold, arbitrary duty-cycle for representing any specific SNs can be generated with the accuracy of 96%. In this manner, the implementation cost of the stochastic number generator (SNG) is reduced by avoiding the use of complex frequency-programmable-oscillator and comparator which are exploited in the conventional TBSC circuit. From circuit simulation results, multiplication and addition operations can be retrieved by proposed TBCS circuits with the accuracy of about 97%. The entire circuitry utilizes 210 CMOS transistors, and consumes the energy of 2.5pJ for one computation, which is 14% and 36.7% of conventional TBCS circuit, respectively.
- Armin Alaghi and John P. Hayes. 2013. Survey of stochastic computing. ACM Transactions on Embedded computing systems (TECS), Vol. 12, 2s (2013), 92. Google ScholarDigital Library
- Armin Alaghi, Cheng Li, and John P. Hayes. 2013. Stochastic circuits for real-time image-processing applications. In Design Automation Conference (DAC), 2013 50th ACM/EDAC/IEEE. IEEE, 1--6. Google ScholarDigital Library
- Te-Hsuan Chen and John P. Hayes. 2014. Analyzing and controlling accuracy in stochastic circuits. In Computer Design (ICCD), 2014 32nd IEEE International Conference on. IEEE, 367--373.Google Scholar
- Guoqiang Hang, Xiaohui Hu, Hongli Zhu, and Xiaohu You. 2013. Differential edge-triggered flip-flops using neuron-mos transistors. In 2013 Ninth International Conference on Computational Intelligence and Security. IEEE, 313--317. Google ScholarDigital Library
- John P. Hayes. 2015. Introduction to stochastic computing and its challenges. In Proceedings of the 52nd Annual Design Automation Conference. ACM, 59. Google ScholarDigital Library
- Hideyuki Ichihara, Tatsuyoshi Sugino, Syota Ishii, Tsuyoshi Iwagaki, and Tomoo Inoue. 2016. Compact and accurate digital filters based on stochastic computing. IEEE Transactions on Emerging Topics in Computing (2016). Google ScholarDigital Library
- Siting Liu and Jie Han. 2018. Toward Energy-Efficient Stochastic Circuits Using Parallel Sobol Sequences. IEEE Transactions on Very Large Scale Integration (VLSI) Systems (2018). Google ScholarDigital Library
- Chengguang Ma, Shunan Zhong, and Hua Dang. 2012. Understanding variance propagation in stochastic computing systems. In Computer Design (ICCD), 2012 IEEE 30th International Conference on. IEEE, 213--218. Google ScholarDigital Library
- M. Hassan Najafi, Shiva Jamali-Zavareh, David J. Lilja, Marc D. Riedel, Kia Bazargan, and Ramesh Harjani. 2017. An Overview of Time-Based Computing with Stochastic Constructs. IEEE Micro, Vol. 37, 6 (2017), 62--71.Google ScholarCross Ref
- M. Hassan Najafi, Shiva Jamali-Zavareh, David J. Lilja, Marc D. Riedel, Kia Bazargan, and Ramesh Harjani. 2017. Time-encoded values for highly efficient stochastic circuits. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, Vol. 25, 5 (2017), 1644--1657. Google ScholarDigital Library
- Florian Neugebauer, Ilia Polian, and John P. Hayes. 2017. Building a Better Random Number Generator for Stochastic Computing. In Digital System Design (DSD), 2017 Euromicro Conference on. IEEE, 1--8.Google Scholar
- Weikang Qian, Xin Li, Marc D. Riedel, Kia Bazargan, and David J. Lilja. 2011. An architecture for fault-tolerant computation with stochastic logic. IEEE Trans. Comput., Vol. 60, 1 (2011), 93--105. Google ScholarDigital Library
- Ao Ren, Zhe Li, Caiwen Ding, Qinru Qiu, Yanzhi Wang, Ji Li, Xuehai Qian, and Bo Yuan. 2017. Sc-dcnn: Highly-scalable deep convolutional neural network using stochastic computing. ACM SIGOPS Operating Systems Review, Vol. 51, 2 (2017), 405--418.Google ScholarCross Ref
- Tadashi Shibata and Tadahiro Ohmi. 1992. A functional MOS transistor featuring gate-level weighted sum and threshold operations. IEEE Transactions on Electron devices, Vol. 39, 6 (1992), 1444--1455.Google ScholarCross Ref
- Paishun Ting and John P. Hayes. 2017. On the role of sequential circuits in stochastic computing. In Proceedings of the on Great Lakes Symposium on VLSI 2017. ACM, 475--478. Google ScholarDigital Library
- Ran Wang, Jie Han, Bruce F. Cockburn, and Duncan G. Elliott. 2016. Stochastic circuit design and performance evaluation of vector quantization for different error measures. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, Vol. 24, 10 (2016), 3169--3183. Google ScholarDigital Library
- Aidyn Zhakatayev, Sugil Lee, Hyeonuk Sim, and Jongeun Lee. 2018. Sign-magnitude SC: getting 10X accuracy for free in stochastic computing for deep neural networks. In 2018 55th ACM/ESDA/IEEE Design Automation Conference (DAC). IEEE, 1--6. Google ScholarDigital Library
- Renyuan Zhang and Mineo Kaneko. 2014. A feasibility study on robust programmable delay element design based on neuron-mos mechanism. In Proceedings of the 24th edition of the great lakes symposium on VLSI. ACM, 21--26. Google ScholarDigital Library
- Renyuan Zhang and Mineo Kaneko. 2015. A feasibility study of quaternary FPGA designs by implementing Neuron-MOS mechanism. In 2015 IEEE International Symposium on Circuits and Systems (ISCAS). IEEE, 942--945.Google ScholarCross Ref
Index Terms
- An Efficient Time-based Stochastic Computing Circuitry Employing Neuron-MOS
Recommendations
1 MHz---3.5 GHz, wide range input duty 50% output duty cycle corrector
A duty cycle corrector (DCC) is proposed in this paper. By directly controlling the duty cycle in the clock distribution path, it can work at a frequency as high as 3.5 GHz. The DCC adopts the continuous-time integrator as the duty cycle detector. The ...
Neuron-MOS Parallel Search Hardware for Real-Time Signal Processing
Selected papers from the 1st Analog VLSI WorkshopCircuit architecture for parallel data processing directly carried out on the hardware have been developed based on a high-functionality transistor, neuron MOSFET (neuMOS or νMOS for short). In the νMOS data sorting circuit, multiple analog input data ...
A feasibility study on robust programmable delay element design based on neuron-MOS mechanism
GLSVLSI '14: Proceedings of the 24th edition of the great lakes symposium on VLSIThe feasibility of programmable delay elements (PDEs) design based on Neuron-MOS mechanism is investigated in this work. By applying the capacitor coupling technology, the charging/discharging current of a clock buffer can be digitally programmed to ...
Comments