11.2 Variability and Robustness for Emerging Technologies

Printer-friendly version PDF version

Date: Thursday 12 March 2015
Time: 14:00 - 15:30
Location / Room: Belle Etoile

Chair:
Edith Beigne, CEA-Leti, FR

Co-Chair:
Andy Tyrrell, University of York, GB

Issues relating to smaller device sizing and novel technologies require more consideration of variability when designing systems and the related robustness of such systems. The first paper in this session considers the modelling of resistive switching random access memory and used in a number of designs to illustrate various properties and characteristics of such devices, including speed-power performance, variability and a neuromorphic computing application. The second paper introduces methods for improving the performance of Spin-Torque Transfer RAM (STTRAM) to reduce worst-case write latency and improve power over more global methods. The third paper proposes a joint optimization of the reliability at device circuit and architecture level. The device level is mainly considered through the energy barrier, circuit level through the transistor controlling the writing current and the architecture level through the error code correction scheme complexity. The final paper in the session compare sub-10nm node TFETs against the projected FinFETs of the same node at 0.25V in both inverter chains and in synthesizing a LEON3 processor.

TimeLabelPresentation Title
Authors
14:0011.2.1VARIATION-AWARE, RELIABILITY-EMPHASIZED DESIGN AND OPTIMIZATION OF RRAM USING SPICE MODEL
Speakers:
Haitong Li1, Zizhen Jiang2, Peng Huang3, Yi Wu2, Hong-Yu Chen2, Bin Gao3, Xiaoyan Liu3, Jinfeng Kang3 and H.-S. Philip Wong2
1Stanford University & Peking University, ; 2Stanford University, ; 3Peking University,
Abstract
Resistive switching random access memory (RRAM) is a leading candidate for next-generation nonvolatile and storage-class memories and monolithic integration of logic with memory interleaved in multiple layers. To meet the increasing need of device-circuit-system co-design and optimization for applications from digital memory systems to brain-inspired computing systems, a SPICE model of RRAM that can reproduce essential device physics in a circuit simulation environment is required. In this work, we develop an RRAM SPICE model that can capture all the essential device characteristics such as stochastic switching behaviors, multi-level cell, switching voltage variations, and resistance distributions. The model is verified and calibrated by a variety of electrical measurements on ~10 nm RRAMs. The model is applied to explore a wide range of applications including: 1) variation-aware design; 2) reliability-emphasized design; 3) speed-power assessment; 4) array architecture optimization; and 5) neuromorphic computing. This experimentally verified design tool not only enables system design that includes the complete suite of RRAM device features, but also provides solutions for system optimization that capitalize on device/circuit interaction.

Download Paper (PDF; Only available from the DATE venue WiFi)
14:3011.2.2IMPACT OF PROCESS-VARIATIONS IN STTRAM AND ADAPTIVE BOOSTING FOR ROBUSTNESS
Speakers:
Seyedhamidreza Motaman, Swaroop Ghosh and Nitin Rathi, University of South Florida, US
Abstract
Spin-Torque Transfer Random Access Memory (STTRAM) is a promising technology for high density on-chip cache due to low standby power. Additionally, it offers fast access time, good endurance and retention. However, it suffers from poor write latency and write power. Additionally we observe that process variation can result in large spread in write and read latency variations. The performance of conventionally designed STTRAM cache can degrade as much as 10% due to process variations. We propose a novel and adaptive write current boosting to address this issue. The bits experiencing worst-case write latency are fixed through write current boosting. Simulations show 80% power improvement compared to boosting all bit-cells and 13% performance improvement compared to worst case latency due to process variation over a wide range of PARSEC benchmarks.

Download Paper (PDF; Only available from the DATE venue WiFi)
15:0011.2.3DEVICE/CIRCUIT/ARCHITECTURE CO-DESIGN OF RELIABLE STT-MRAM
Speakers:
Zoha Pajouhi, Xuanyao Fong and Kaushik Roy, Purdue University, US
Abstract
Spin transfer torque magnetic random access memory (STT-MRAM), using magnetic tunnel junctions (MTJ) has garnered significant attention in the research community due to its immense potential for on-chip, high-density and non-volatile memory. However, process variations may significantly impact the achievable yield in STT-MRAM. To this end, several yield enhancement techniques that improve STT-MRAM failures at the bit-cell, and at the architecture level of design abstraction have been proposed in the literature. However, these techniques may lead to a suboptimal design because they do not consider the impact of design choices at every level of design abstraction. In this paper, we propose a unified device-circuit-architecture co-design framework to optimize and enhance the yield of STT-MRAM. We studied the interaction between device parameters (viz. energy barrier height) and bit-cell level parameters (viz. transistor width), together with different Error Correcting Codes (ECC) to optimize the robustness and energy efficiency of STT-MRAM cache. The advantages of our proposed approach to STT-MRAM design are explored at the 32nm technology node. We show that for a target yield of 500 Defects Per Million (DPM) for an example array with 64-bit word length, our proposed ap-proach with realistic parameters can save up to 15% and 13% in cell area and total power consumption, respectively, in compari-son with a design that does not use any array level yield en-hancement technique.

Download Paper (PDF; Only available from the DATE venue WiFi)
15:1511.2.4SUB-10 NM FINFETS AND TUNNEL-FETS: FROM DEVICES TO SYSTEMS
Speakers:
Ankit Sharma, Arun Goud Akkala and Kaushik Roy, Purdue University, US
Abstract
In this paper, a detailed device/circuit/system level assessment of sub-10nm GaSb-InAs Tunneling Field Effect Transistors (TFET) versus Silicon FinFETs operating at near-threshold voltages is reported. A source underlapped GaSb-InAs TFET is used to achieve lower subthreshold swings than previously reported TFETs and an analytical justification is provided to explain the observed improvement. Through atomistic, 2D ballistic simulations using self-consistently, coupled Non-equilibrium Green's Function (NEGF)-Poisson approach, GaSb-InAs TFET and Silicon FinFET device characteristics are derived from which compact models are extracted for SPICE simulations. Circuit simulations of a 6-stage inverter chain show that sub-10nm underlapped TFETs are especially suited for near-threshold computing because of their ability to achieve higher throughput while consuming ~100x lower power compared to Si FinFETs. To analyze the suitability of sub-10 nm TFETs for medium-throughput and ultra-low power applications in future very large scale integrated designs, a LEON3 processor is synthesized at VDD=0.25V. The impact of interconnect parasitics on the performance of TFETs is considered by studying the power-performance of the LEON3 under varying wire-load conditions. Under moderate interconnect parasitics, TFETs-based processor is shown to exhibit more than 50% power reduction compared to FinFETs.

Download Paper (PDF; Only available from the DATE venue WiFi)
15:30IP5-10, 460SPINTASTIC: SPIN-BASED STOCHASTIC LOGIC FOR ENERGY-EFFICIENT COMPUTING
Speakers:
Rangharajan Venkatesan1, Swagath Venkataramani2, Xuanyao Fong2, Kaushik Roy2 and Anand Raghunathan2
1NVIDIA Corporation, US; 2Purdue University, US
Abstract
Spintronics is one of the leading technologies under consideration for the post-CMOS era. While spintronic memories have demonstrated great promise due to their density, non-volatility and low leakage, efforts to realize spintronic logic have been much less fruitful. Recent studies project the performance and energy efficiency of spintronic logic to be considerably inferior to CMOS. In this work, we explore Stochastic Computing (SC) as a new direction for the realization of energy-efficient logic using spintronic devices. We establish the synergy between stochastic computing and spintronics by demonstrating that (i) the peripheral circuits required for SC to convert to/from stochastic domains, which incur significant energy overheads in CMOS, can be efficiently realized by exploiting the characteristics of spintronic devices, and (ii) the low logic complexity and fine-grained parallelism in SC circuits can be leveraged to alleviate the shortcomings of spintronic logic. We propose SPINTASTIC, a new design approach in which all the components of stochastic circuits — stochastic number generators, stochastic arithmetic units, and stochastic-to-binary converters — are realized using spintronic devices. Our experiments on a range of benchmarks from different application domains demonstrate that SPINTASTIC achieves 2.8X improvement in energy over CMOS stochastic implementations and 1.9X over a CMOS binary baseline.

Download Paper (PDF; Only available from the DATE venue WiFi)
15:31IP5-11, 520LEAKAGE POWER REDUCTION FOR DEEPLY-SCALED FINFET CIRCUITS OPERATING IN MULTIPLE VOLTAGE REGIMES USING FINE-GRAINED GATE-LENGTH BIASING TECHNIQUE
Speakers:
Ji Li, Qing Xie, Yanzhi Wang, Shahin Nazarian and Massoud Pedram, University of Southern California, US
Abstract
With the aggressive downscaling of the process technologies and importance of battery-powered systems, reducing leakage power consumption has become one of the most crucial design challenges for IC designers. This paper presents a device-circuit cross-layer framework to utilize fine-grained gate-length biased FinFETs for circuit leakage power reduction in the near- and super-threshold operation regimes. The impacts of Gate-Length Biasing (GLB) on circuit speed and leakage power are first studied using one of the most advanced technology nodes - a 7nm FinFET technology. Then multiple standard cell libraries using different leakage reduction techniques, such as GLB and Dual-VT, are built in multiple operating regimes at this technology node. It is demonstrated that, compared to Dual-VT, GLB is a more suitable technique for the advanced 7nm FinFET technology due to its capability of delivering a finer-grained trade-off between the leakage power and circuit speed, not to mention the lower manufacturing cost. The circuit synthesis results of a variety of ISCAS benchmark circuits using the presented GLB 7nm FinFET cell libraries show up to 70% leakage improvement with zero degradation in circuit speed in the near- and super-threshold regimes, respectively, compared to the standard 7nm FinFET cell library.

Download Paper (PDF; Only available from the DATE venue WiFi)
15:30End of session
Coffee Break in Exhibition Area

Coffee Break in Exhibition Area

On all conference days (Tuesday to Thursday), coffee and tea will be served during the coffee breaks at the below-mentioned times in the exhibition area.

Lunch Break

On Tuesday and Wednesday, lunch boxes will be served in front of the session room Salle Oisans and in the exhibition area for fully registered delegates (a voucher will be given upon registration on-site). On Thursday, lunch will be served in Room Les Ecrins (for fully registered conference delegates only).

Tuesday, March 10, 2015

Coffee Break 10:30 - 11:30

Lunch Break 13:00 - 14:30; Keynote session from 13:20 - 14:20 (Room Oisans) sponsored by Mentor Graphics

Coffee Break 16:00 - 17:00

Wednesday, March 11, 2015

Coffee Break 10:00 - 11:00

Lunch Break 12:30 - 14:30, Keynote lectures from 12:50 - 14:20 (Room Oisans)

Coffee Break 16:00 - 17:00

Thursday, March 12, 2015

Coffee Break 10:00 - 11:00

Lunch Break 12:30 - 14:00, Keynote lecture from 13:20 - 13:50

Coffee Break 15:30 - 16:00