[paper] SPICE-level Crossbar-array Circuit Simulator

Fan Zhang¹ and Miao Hu² 

CCCS: Customized SPICE-level Crossbar-array Circuit Simulator

for In-Memory Computing

IEEE/ACM International Conference on Computer-Aided Design

(ICCAD ’20) November 2– 5, 2020, Virtual Event, USA. 

ACM, New York, NY, USA, 8 pages.

DOI: 10.1145/3400302.3415627

1Arizona State University Tempe, Arizona
2Binghamton University Binghamton, New York

ABSTRACT: Resistive crossbar arrays are known for their unique structure to implement analog in-memory vector-matrix-multiplications (VMM). However, general-purpose circuit simulators, such as HSPICE and HSIM, are too slow for large scale crossbar array simulations with consideration of circuit parasitics. Although there are some specific simulators designed for crossbar arrays, they mainly focus on area/power/delay estimation rather than accurate SPICE-level simulation, thus could not model its functionality on analog in-memory computing. In this paper, we firstly give a SPICE-level modeling of resistive crossbar array with consideration of circuit parasitics in MATLAB. We also propose efficient methods to further speedup simulations by model simplifications. Last but not least, ResNet-20 on CIFAR-10 is applied to demonstrate the work. With the proposed model simplification methods, simulation speed can be improved by ~31X with tolerable errors, and more than 5X speedup is achieved on ResNet-20 while the accuracy drop is 6%.

Figure: Implement the ResNet on the crossbar with sub-block optimization. 

RELATED WORK: Other than general-purpose circuit simulators, specific simulation platforms have been proposed for crossbar-based application analysis; examples include: 

[MNSIM] L. Xia, B. Li, T. Tang, P. Gu, X. Yin, W. Huangfu, P. Chen, S. Yu, Y. Cao, Y. Wang, Y. Xie, and H. Yang. MNSIM: Simulation platform for memristor-based neuromorphic computing system. In 2016 Design, Automation Test in Europe Conference Exhibition (DATE). 469–474.
[NeuroSim] P. Chen, X. Peng, and S. Yu. 2018. NeuroSim: A Circuit-Level Macro Model for Benchmarking Neuro-Inspired Architectures in Online Learning. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems 37, 12 (Dec 2018), 3067–3080.

#Urban Explorers Reveal A #Treasure Trove Of #Soviet #Computing Power

#Urban Explorers Reveal A #Treasure Trove Of #Soviet #Computing Power https://t.co/5dLGxTSMRF#semi 

— Wladek Grabinski (@wladek60) November 30, 2020

---

from Twitter https://twitter.com/wladek60

November 30, 2020 at 07:55PM
via IFTTT

[paper] The advantages of p-GaN channel/Al2O3 gate insulator

Maria Ruzzarin,1, Carlo De Santi,1 Feng Yu,2 Muhammad Fahlesa Fatahilah,2 Klaas Strempel,2 Hutomo Suryo Wasisto,2 Andreas Waag,2 Gaudenzio Meneghesso,1 Enrico Zanoni,1

and Matteo Meneghini1

Highly stable threshold voltage in GaN nanowire FETs: The advantages of p-GaN channel/Al2O3 gate insulator

Appl. Phys. Lett. 117, 203501 (2020); 

DOI: 10.1063/5.0027922

Published Online: 16 November 2020

1 Department of Information Engineering, University of Padova, via Gradenigo 6/b, 35131 Padova, Italy
2 Institute of Semiconductor Technology (IHT) and Laboratory for Emerging Nanometrology (LENA), Technische Universitat Braunschweig, Langer Kamp 6a/b, 38106 Braunschweig, Germany

Abstract: We present an extensive investigation of the charge-trapping processes in vertical GaN nanowire FETs with a gate-all-around structure. Two sets of devices were investigated: Gen1 samples have unipolar (n-type) epitaxy, whereas Gen2 samples have a p-doped channel and an n-p-n gate stack. From experimental results, we demonstrate the superior performance of the transistor structure with a p-GaN channel/Al2O3 gate insulator in terms of dc performance. In addition, we demonstrate that Gen2 devices have highly stable threshold voltage, thus representing ideal devices for power electronic applications. Insight into the trapping processes in the two generations of devices was obtained by modeling the threshold voltage variations via differential rate equations.

Fig. a) The p-channel device (Gen2) comprises a 2.5 lm n-GaN buffer layer, a 0.5 lm p-GaN channel layer, 0.73 lm n-GaN and 0.5 lm n p-GaN as the top layer, and 25 nm-Al2O3 as the gate dielectric.

b) SEM images of a nanowire of the p-channel device (Gen2) and bird’s-eye view of vertically aligned n-p-n GaN nanowire (NW) arrays with top contacts.

Aknowledgement: This work was supported in part by NoveGaN (Univ. of Padova) through the STARS CoG Grants call. Ack prog. Eccellenza. This research was partly performed within project INTERNET OF THINGS: SVILUPPI METODOLOGICI, TECNOLOGICI E APPLICATIVI and co-funded (2018–2022) by the Italian Ministry of Education, Universities and Research (MIUR) under the aegis of the “Fondo per il finanziamento dei dipartimenti universitari di eccellenza” initiative (Law 232/2016). Financial support from the German Research Foundation (DFG) of 3D GaN project and the Lower Saxony Ministry of Science and Culture (N-MWK) of LENA-OptoSense group is highly acknowledged for the development of vertical GaN nanowire FETs.

#top10 Hottest Semiconductor #Startups Of #2020

#top10 Hottest Semiconductor #Startups Of #2020 https://t.co/wIJrk3NU6b #semi

— Wladek Grabinski (@wladek60) November 29, 2020

---

from Twitter https://twitter.com/wladek60

November 29, 2020 at 08:50PM
via IFTTT

[paper] How Objective is Peer Review?

By Hannah Bast 

November 18, 2020

In 2014, the organizers of the Conference on Neural Information Processing Systems (NeurIPS, then still called NIPS) made an interesting experiment.1 They split their program committee (PC) in two and let each half independently review a bit more than half of the submissions. That way, 10% of all submissions (166 papers) were reviewed by two independent PCs. The aimed at acceptance rate per PC was 23%. The result of the experiment was that among these 166 papers, the set of accepted papers from the two PCs overlapped by only 43%. That is, more than half of the papers accepted by one PC were rejected by the other. This led to a passionate flare-up of the old debate of how effective or random peer-reviewing really is and what we should do about it. [read more...]

My bottom line: The reputation of the peer review process is tarnished. Let us work on this with the same love and attention we give to our favorite research problems. Let us do more experiments to gain insights that help us make the process more fair and regain some trust. And let us create powerful incentives, so that whatever we already know is good is actually implemented and carried over from one PC to the next.

References: 

1 https://cacm.acm.org/blogs/blog-cacm/181996-the-nips-experiment provides a short description of the NIPS experiment and various links to further analyses and discussions.

2 https://github.com/ad-freiburg/esa2018-experiment

3 There are other experiments, like the single-blind vs. double-blind experiment at WSDM'17, which investigated a particular aspect of the reviewing process: https://arxiv.org/abs/1702.00502

Hannah Bast is a professor of computer science at the University of Freiburg, Germany. Before that, she was working at Google, developing the public transit routing algorithm for Google Maps. Right after the ESA experiment, she became Dean of the Faculty of Engineering in Freiburg and a member of the Enquete Commission for Artificial Intelligence of the German parliament (Bundestag). That's why it took her two years to write this blog post.