Search

For sub-7-nm technology nodes, the gate-all-around (GAA) nanowire-based device structure is a strong candidate to sustain scaling according to Moore’s Law. For the first time, the performance of two GAA device options—lateral FET (LFET) and vertical FET (VFET)—is benchmarked and analyzed at the system level using an ARM core processor, based on realistic compact device models at the 5-nm technology node. Tradeoffs among energy, frequency, leakage, and area are evaluated by a multi- Vth optimization flow. A variety of relevant device configurations, including various number of fins, nanowires, and nanowire stacks, are explored. The results demonstrate that an LFET GAA core has a larger maximum frequency than its VFET counterpart because the channel stress that can be created in the LFETs results in a larger ON current. For fast timing targets, the LFET cores are therefore superior. However, for slow timing targets (e.g., 5 ns), the VFET cores with three nanowires offer a 7% area reduction and a 20% energy saving compared with the LFET cores with 2fin/2stack at the same leakage power. [ABSTRACT FROM PUBLISHER]

Based on realistic circuit- and system-level simulations, graphene interconnects are analyzed in terms of multiple material properties, such as the mean free path (MFP), the contact resistance, and the edge roughness. The benchmarking results indicate that the advantage of using graphene interconnects occurs only under certain circumstances. The device-level parameters, including the supply and threshold voltages, and the circuit-level parameters, including the wire length and width, have large impacts on both the delay and energy-delay product (EDP). At the circuit level, one representative circuit, a 32-bit adder, is investigated, where up to 40% and 70% improvements in delay and EDP are observed. At the system-level analysis, an ARM Cortex-M0 processor is synthesized, and placement and routing are performed. After replacing copper interconnects with multilayer graphene interconnects, up to 15% and 22% performance improvements in clock frequency and EDP have been observed. It is also demonstrated that the benefits of using graphene for the ARM core processor are strongly dependent on the quality of the graphene, such as the MFP and the edge roughness. [ABSTRACT FROM AUTHOR]

Abstract: With the growth of the embedded devices consumer market, power efficient hardware is needed. Therefore power-aware architectural exploration is one of the most crucial design steps. For such an exploration procedure, it is important to accurately model the power consumption of all main components of the embedded system. Registers and register files are one of the highest power consumers of any programmable processor, but there is a lack of accurate and publicly available models. This paper provides such a power model for standard cell based register files for 130 and 90nm technologies. The proposed model provides dynamic power, leakage power, area and timing information for register files in terms of key parameters like width, depth, activity, ports, and capacitive loading. It is shown that current models capture neither correct absolute nor relative trends present in register files. It is shown that some key, but often neglected parameters like switching activity, load have a larger influence in some particular sizes of the register files than others. Therefore, using the Empire model, accurate architectural exploration is possible. [Copyright &y& Elsevier]

Reduced energy consumption is one of the most important design goals for embedded application domains like wireless communication, multimedia, and biomedical applications. The instruction memory hierarchy has been proven to be one of the most power hungry parts of the system. This paper introduces an architectural enhancement for the instruction memory to reduce energy consumption and improve performance. The proposed distributed instruction memory organization requires minimal hardware overhead and supports the execution of multiple incompatible loops in parallel in a uniprocessor system. We present different methods to implement the loop controller architecture, compare them, and show that distributing the instruction memory helps to reduce the interconnect cost as well. This architecture enhancement can reduce the energy consumed in the instruction memory hierarchy by 59 percent and improve the performance by 22 percent compared to hardware-based enhanced-SMT-based architectures. [ABSTRACT FROM AUTHOR]

New applications demand very high processing power when run on embedded systems. Very Long Instruction Word (VLIW) architectures have emerged as a promising alternative to provide such processing capabilities under the given energy budget. However, in this new VLIW-based architectures, the register file is a very critical contributor to the overall power consumption and new approaches have to be proposed to reduce its power while preserving system performance. In this paper, we propose a novel joint hardware–software approach that reduces the leakage energy in the register files of these embedded VLIW architectures. This approach relies upon an energy-aware register assignment method and a hardware support that creates sub-banks in the global register file that can be switched on/off at run time. Our results indicate energy savings in the register file, after considering the overhead of the added extra hardware, up to 50% for modern multimedia embedded applications without performance degradation. We illustrate this approach using real-life applications running on these processors. We also illustrate the tradeoff between the area overhead vs. the gains in the leakage energy for the different strategies. [Copyright &y& Elsevier]

An exploratory modeling methodology is presented for estimating power noise in advanced technology nodes. The models are evaluated for 14, 10, and 7 nm FinFET technologies to assess the impact on performance. The power noise is composed of three parts, noise related to the global power grids, via stacks, and local power rails, based on the hierarchical nature of power distribution networks. In 14 nm technology, the global power noise dominates the total power noise. The power noise is lower and more evenly distributed in 10 nm technology. 7 nm technology is shown to be more sensitive to local power noise. To decrease the global power noise, extra metal layers are added to the global power grid. A 75% reduction in global power noise is observed in 14 nm technology. Stripes between local track rails are evaluated to reduce the local power noise, exhibiting up to 57% improvement in local power noise at the 7 nm technology node. As a promising alternative material for power network interconnects, few layer graphene is shown to exhibit good potential for reducing local power noise. The effects of different scaling scenarios of the local power rails on power noise are also discussed. [ABSTRACT FROM AUTHOR]

A novel sub-threshold 9 T Static Random Access Memory (SRAM) cell designed and simulated in 14-nm FinFET technology is proposed in this paper. The proposed 9 T-SRAM cell offers an improved access time in comparison to the 8 T-SRAM cell. Furthermore, an assist circuit is proposed by which the leakage current of the proposed SRAM cell is reduced by 20% when holding '0' and an equal leakage current during hold '1' in comparison to the 8 T-SRAM cell. The proposed circuit improves the access time by 40% in comparison to the 8 T-SRAM cell without any degradation in write and read noise margins, as well. The maximum operating frequency of the proposed SRAM cell is 1.53 MHz at VDD = 270 mV. Copyright © 2016 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

In this paper, we benchmark the performance of monolayer and bilayer MoS2 FETs (MFETs) against various multigate (MuG) Si-FET options, such as FinFETs and lateral and vertical nanowire FETs, for a 5-nm node and beyond. We compare the performance metrics of all the device options at the ring-oscillator (RO) level, accounting for not only intrinsic and extrinsic parasitic elements but also interconnects. Using the atomistic two-band ballistic quantum transport simulations, we evaluate ON-current and intrinsic capacitances for MoS2-based devices. Furthermore, we calibrate two-band model currents with more sophisticated full-band diffusive simulations to obtain realistic performance metrics at the circuit level. We show that both the intrinsic and parasitic capacitances of a single-gate MFET are lesser than those of a double-gate (DG) MFET, resulting in 13% lesser energy consumption. A DG bilayer (DGBL) MFET shows the best performance among different MFETs. In comparison toMuG FETs, the DGBL MFET offers not only lower energy consumption but also 35%–45% lower speed. In the end, to meet the target performance, we evaluate the impact of the device current, contact resistance, and back-end-of-the-line load on the speed of RO with the DGBL MFET. [ABSTRACT FROM PUBLISHER]

For the first time, we present complete device threshold voltage (VT)-targeting methodology for FinFET SRAM in 10-nm technology, considering capacitance due to metal pattering and device variability to set target read current for different variants of SRAM architecture to determine technology high-density (HD) SRAM cell. The VT-targeting methodology brings into play the worst case read and write margins available for SRAM cell to determine nominal device VT by tuning the work function of metal gate. Analysis shows that for minimum leakage current, 112 SRAM cell is optimum, whereas for the same area of 0.0546~\mu \mathrm{m}^{\mathrm {2}} with 50% higher leakage, 122 SRAM outperform by 5% and 20% improved read and write margins, respectively. The 122 SRAM as HD cell reduces the cost of the technology by sharing P-channel field effect transistor (PFET) and N-channel field effect transistor (NFET) VT mask with the high threshold voltage logic devices, whereas the 112 SRAM device shares only NFET VT mask. The 111 SRAM can achieve target performance at lesser area of 0.048~\mu \mathrmm^\mathrm 2 by compromising read stability, which will result in lower yield. At 64-nm pitch, litho-etch litho-etch (LELE) double-patterned gate impacts device performance and alleviates variability; hence the read margin of SRAM cell should consider an additional 1\sigma \mathrm {\mathbf {rsnm}} margin to retain the same yield in 10-nm-technology era. [ABSTRACT FROM AUTHOR]

Abstract: In deeply scaled CMOS technologies, Bias Temperature Instability (BTI) is one of the most critical degradation mechanisms impacting the device reliability. This study presents the BTI evaluation of gates covering both the PMOS and NMOS degradation in a workload dependent, atomistic trap-based, stochastic BTI model. The gate propagation delay depends on the gate intrinsic delay, the input signal characteristics, and the output load. In this paper, the impact of (1) duty factor, (2) periodic clock-based and non-periodic random input sequences, (3) gate, and (4) drive strength to the BTI degradation are investigated. Statistical studies show a mean degradation of 3% and a worst-case of 27%. Moreover, the near-critical paths with lower drive strength cells are 3.7× more susceptible to BTI degradation than the critical paths with higher drive strength cells. Next, the relative degradations of the propagation delays for the well-known gates (i.e. INV, NAND, NOR, AOI) are presented. Under the same stress stimuli, degradations of the gate propagation delays differ by 4.5×. [Copyright &y& Elsevier]

Next generation mobile wireless systems (4G) support a wide range of communication protocols and services, thus opening new design challenges. The desired flexibility requires an effective utilization of system resources. In this article, we introduce the concept of system scenarios in wireless baseband engine signal processing optimization and in digital front-end power optimization. The scenario methodology classifies the system behavior from a cost perspective and provides the necessary information for an effective system tuning. We propose improvements for the clustering of the system executions into scenarios and the detection of scenarios at run time achieving a better trade-off between cost estimation accuracy and detection overhead. The first case study of the paper, using the WLAN communication protocol, demonstrates the accurate prediction of the execution time of each block of bits, which on average is 92 % shorter than the worst case allowing us to use the remaining time for the optimization of specifications like power consumption. In the second case study, we concentrate on the efficient signal power management during a WLAN transmission reducing the total energy consumption 50-94 % based on the throughput utilization. [ABSTRACT FROM AUTHOR]

This paper presents an accurate technique to extensively analyze the impact of time-zero (i.e., global and local variation) and time-dependent (i.e., voltage, temperature, workload, and aging) variation on the offset voltage specification of a memory sense amplifier design using 45 nm predictive technology model (PTM) high performance library. The results show that increasing the supply voltage both for time-zero and time-dependent reduces the offset voltage specification marginally, irrespective of the process corners. In contrast, the offset voltage specification is very sensitive to the temperature and the workload, i.e., the applied voltage patterns. The results also show that a balanced workload results in a significantly lower offset voltage specification. The above results can be used to estimate the required offset voltage accurately for a given lifetime, and operational conditions such as workload, temperature, and voltage; hence, enable the designer to take appropriate measures for a high quality, robust, optimal and reliable design. [ABSTRACT FROM AUTHOR]

Standard-cell design, technology choices, and place and route (P&R) efficiency are deeply interrelated in CMOS technology nodes below 10 nm, where lower number of tracks cells and higher pin densities pose increasingly challenging problems to the router in terms of congestion and pin accessibility. To evaluate and downselect the best solutions, a holistic design-technology co-optimization approach leveraging state-of-theart P&R tools is thus necessary. We adopt such an approach using the imec N7 technology platform, with contacted poly pitch of 42 nm and tightest metal pitch of 32 nm, by comparing post P&R area of an IP block for different standard cell configurations, technology options, and cell height. Keeping the technology node and the set of ground rules unchanged, we demonstrate that a careful combination of these solutions can enable area gains of up to 50%, comparable with the area benefits of migrating to another node. We further demonstrate that these area benefits can be achieved at isoperformance with >20% reduced power. As at the end of the CMOS roadmap, conventional scaling enacted through pitch reduction is made more and more challenging by constraints imposed by lithography limits, material resistivity, manufacturability, and ultimately wafer cost, the approach shown herein offers a valid, attractive, and low-cost alternative. [ABSTRACT FROM AUTHOR]

Self-aligned multiple patterning, due to its low overlay error, has emerged as the leading option for 1-D gridded back-end-of-line (BEOL) in sub-14-nm nodes. To form actual routing patterns from a uniform “sea of wires,” cut masks are needed for line-end cutting or realization of space between routing segments. The line-end cutting results in nonfunctional (i.e., dummy fill) patterns that change wire capacitance, and hence design timing and power. Therefore, to remove such dummy fill patterns, extra 2-D block masks are used. However, 2-D block masks cannot remove arbitrary dummy fill patterns, due to design rule constraints on the block mask shapes. In this paper, we address the timing-aware optimization of 2-D block mask layouts under various sets of mask rules that are derived from mask patterning technology options (e.g., 193i and 193d) for foundry 7-/5-nm (N7/N5) BEOL. Our central contribution is a mixed integer linear programming (MILP) optimization that minimizes timing impact due to dummy metal segments while satisfying block mask rules and metal density constraints. We also propose a distributed optimization flow to improve the scalability. With our optimizer, we recover up to 84% of the worst negative slack impact from dummy segments, with up to 64% dummy removal rate. We further extend our MILP to a co-optimization of cut and block masks. This paper gives new insights into fundamental limits of benefit from emerging cut and block mask technology options. [ABSTRACT FROM AUTHOR]

In this paper, lateral gate-all-around nano-sheet transistors (NSH-FETs) are explored from intrinsic performance to dc and ring oscillator (RO) benchmark compared with FinFETs and nanowire transistors (NW-FETs) for sub-7-nm node. The band structure calculated technology computer aided design results show comparable intrinsic performance to FinFETs at same channel cross section. On top of that, dc and RO are evaluated by taking into account electrostatics, parasitic components, and layout configurations. The NSH-FETs show an advantage in drive current with the NSH width but their RO performance is limited by the device capacitance. The multiple narrow NSH-FET shows ~5% higher drive current compared to the NW-FET at similar subthreshold swing, allowing heavier capacitive loaded circuit. In addition, NSH-FETs can provide the device design freedom from aggressive fin pitch scaling. [ABSTRACT FROM PUBLISHER]

We investigate the impact of wire geometry on the resistance, capacitance, and RC delay of Cu/low- $k$ damascene interconnects for fixed line-to-line pitch. The resistance is computed by applying a semiempirical resistivity model, calibrated to Cu damascene wires, integrated with a Ru-based liner, currently investigated for the 7 nm logic technology node. The capacitance is simulated by means of a 2D field solver (Raphael) by Synopsys. The impact of line dimensions is analyzed for the case of 32 nm pitch interconnects, which are representative of the 7 nm logic technology node. We show that for aspect ratios greater than 1, the resistance is more sensitive to variations of the line width rather than of the line height, because of the higher surface scattering induced by the sidewall interfaces, which are closer to each other compared with the top and bottom interfaces. For capacitance, low- $k$ sidewall damage exacerbates capacitance sensitivity to line dimensions and, for typical interconnect schemes, the impact of line width variations dominates over variations of the line height. We demonstrate that for a given pitch and dielectric stack height, the RC delay can be significantly reduced by targeting wider and deeper damascene trenches, that is, by trading capacitance for resistance, and that an optimal wire geometry for RC delay minimization exists. In addition, we show that a given RC delay can be achieved with several geometries and, therefore, $R$ and $C$ pairs, which represents a useful degree of freedom for designers to optimize system-level performance. As an application, we analyze a possible 7 nm technology scenario and show that wide and deep damascene trenches can mitigate the impact of the increased wire resistance on circuit delay. [ABSTRACT FROM AUTHOR]

In this work, we evaluate the material properties of graphene and assess the potential application of graphene to replace copper wires in Back-End-Of-Line (BEOL) interconnects. Based on circuit and system-level simulations, high restrictions are imposed to graphene with respect to contact resistance and mean free path. Experimentally we evaluate single and multi-layer graphene wires and we measure carrier mean free paths (MFPs) above ~ 110 nm. However, contact engineering will be the key issue for integration of graphene as interconnect. [ABSTRACT FROM AUTHOR]

Conventional junctionless (JL) multi-gate (MuG) field-effect transistors(FETs) require extremely scaled channels to deliver high on-state current with low short-channel effect related leakage. In this letter, using ultra-thin 2D materials (e.g., monolayer MoS2), we present comparison of short-channel effects in JL, and inversion-mode (IM) FETs. We show that JL FETs exhibit better sub-threshold slope (S.S.) and drain-induced-barrier-lowering (DIBL) in comparison to IM FETs due to reduced peak electric field at the junctions. But, threshold voltage (VT) roll-off with channel length downscaling is found to be significantly higher in JL FETs than IM FETs, due to higher source/drain controlled charges (dE/dx) in the channel. Further, we show that although VT roll-off in JL FETs improves by increasing the gate control, i.e., by scaling the oxide, or channel thickness, the sensitivity of threshold voltage on structural parameters is found out to be high. [ABSTRACT FROM AUTHOR]

Standard-cell design and characterization are presented for 7-nm CMOS platform technology targeting low-power and high-performance applications with the tightest contacted poly pitch of 42 nm and a metallization pitch of 32 nm in the FinFET technology. Two standard-cell architectures for 7 nm, a 9-track library and a 7.5-track library have been designed, introducing an extra middle-of-line layer to enable an efficient layout of the 7.5-track cells. The 7.5-track cells are on average smaller than the 9-track cells. With the strict design constraints imposed by self-aligned quadruple patterning and self-aligned double patterning, careful design and technology co-optimization is performed. [ABSTRACT FROM AUTHOR]

Negative Bias Temperature Instability (NBTI) is one of the major time-dependent degradation mechanisms that impact the reliability of advanced deeply scaled CMOS technologies. NBTI can cause workload-dependent shifts on a transistor’s threshold voltage ( V TH ), and performance during its lifetime. This study presents a comparison of the NBTI aging on adder architectures and ring oscillators in the downscaling technology nodes. The first part of the study focuses on the NBTI-induced performance degradation of 32-bit adders (one of the most fundamental block of a processor’s arithmetic logic unit) from the points of architectural topology and workload dependency in the planar technologies (i.e. commercial 28 , 45 , 65 nm nodes), while the second part investigates the energy-delay product degradation of ring oscillators beyond the planar nodes (i.e. research-grade 14 , 10 , 7 nm FinFET technology nodes for several FET channel materials, e.g. Si, SiGe, Ge, InGaAs). Results show the tight coupling between the NBTI aging and the architectural topology, run-time workload, and technology choice. [ABSTRACT FROM AUTHOR]

In this paper, we compare the performances of FinFETs, lateral gate-all-around (GAA) FETs, and vertical GAAFETs (VFETs) at 7-nm node dimensions and beyond. Comparison is done at ring oscillator level accounting not only for front-end of line devices but also for interconnects. It is demonstrated that FinFETs fail to maintain the performance at scaled dimensions, while VFETs demonstrate good scalability and eventually outperform lateral devices both in speed and power consumption. Lateral GAAFETs show better scalability with respect to FinFETs but still consume 35% more energy per switch than VFETs if made under 5-nm node design rules. [ABSTRACT FROM AUTHOR]

This paper describes the implications of bias temperature instability (BTI)-induced time-dependent threshold voltage distributions on the performance and yield estimation of digital circuits. The statistical distributions encompassing both time-zero and time-dependent variability and their correlations are discussed. The impact of using normally distributed threshold voltages, imposed by state-of-the-art design approaches, is contrasted with our defect-centric approach. Extensive Monte Carlo simulation results are shown for static random access memory cell and ring oscillator structures. [ABSTRACT FROM PUBLISHER]