A5001754984- Low-power high-performance VLSI design
- Embedded Systems Design Techniques
- Parallel Computing and Optimization Techniques
- Radiation Effects in Electronics
- Climate Change Policy and Economics
- Advancements in Semiconductor Devices and Circuit Design
- Environmental Impact and Sustainability
- VLSI and FPGA Design Techniques
- Integrated Energy Systems Optimization
- Energy, Environment, and Transportation Policies
- Global Energy and Sustainability Research
- Energy and Environment Impacts
- Semiconductor materials and devices
- Renewable energy and sustainable power systems
- Air Quality and Health Impacts
- Renewable Energy and Sustainability
- Analog and Mixed-Signal Circuit Design
- VLSI and Analog Circuit Testing
- Advanced Data Storage Technologies
University of Stuttgart
2018-2022
Rational (Germany)
2020
Intel (United States)
2008
Georgia Institute of Technology
2001-2007
Embedded Systems (United States)
2006
Major impediments to technology scaling in the nanometer regime include power (or energy) dissipation and "erroneous" behavior induced by process variations noise susceptibility. In this paper, we demonstrate that CMOS devices whose is rendered probabilistic (yielding or PCMOS) can be harnessed for ultra low energy high performance computation. PCMOS are inherently they guaranteed compute correctly with a probability 1= 2< p< 1 thus, design, expected incorrectly (1 p). show yields...
Parameter variations, noise susceptibility, and increasing energy dissipation of cmos devices have been recognized as major challenges in circuit microarchitecture design the nanometer regime. Among these, parameter variations susceptibility are increasingly causing to behave an “unreliable” or “probabilistic” manner. To address these challenges, a shift paradigm from current-day deterministic designs “statistical” is deemed inevitable. respond this need, article, we introduce study entirely...
On its way to a low-carbon society, various transformation pathways for the European energy system are possible. Additionally, improving air quality may benefit from or influence climate change mitigation. In comparative scenario analysis, interactions between decarbonization targets and pollution control as well their implications on studied. For this purpose, model TIMES PanEU is linked with impact assessment EcoSense internalize health costs related pollution. Results indicate that...
An energy system transition in the European Union is required to meet decarbonisation targets determined Paris Agreement. To realise this transition, a reduction greenhouse gas emissions of at least 80% from 1990 levels by 2050 required. However, ambitions are being raised towards carbon neutrality. Such will likely imply deep changes across all sectors EU economies and societies. The impacts on these need be analysed with multi-disciplinary approach, unveil cross-sectoral risks...
Noise immunity and low-energy computing have become limiting factors in the semiconductor roadmap as transistor feature sizes shrink. The subject of our study is probabilistic switch, implemented complementary metal–oxide–semiconductor (CMOS) domain, referred to a CMOS (PCMOS) whose behavior rendered by noise. In conducting this study, we are motivated possibility using such switches realize ultra-low energy computing. Based on PCMOS realized 0.5 0.25 µm processes, present detailed...
The scaling trend of semiconductor devices has raised several issues such as energy consumption and heat dissipation, well the increasing probabilistic behavior devices. Motivated by necessity to consider approaches future designs, CMOS (PCMOS) based computing been proposed. PCMOS are inherently that compute correctly with a probability <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">p</i> . This paper investigates tradeoffs between energy, speed...
With the publication of European Green Deal, Union has committed to reaching carbon neutrality by 2050. The envisaged reductions direct greenhouse gases emissions are seen as technically feasible, but if a wrong path is pursued, significant unintended impacts across borders, sectors, societies and ecosystems may follow. Without insights gained from an impact assessment framework beyond techno-economic perspective, pursuit emission lead counterproductive outcomes in long run. We discuss...
Highly scaled CMOS devices in the nanoscale regime would inevitably exhibit statistical or probabilistic behavior. Such behavior is due to process variations and other perturbations such as noise. Therefore current circuit design methodologies, which depend on existence of deterministic uniform with no consideration for either power consumption behavior, longer be sufficient robust circuits. To help overcome this challenge, we have been characterizing (probabilistic PCMOS devices) at several...
The required decarbonization of the energy system is a complex task, with ambitious targets under Paris Agreement, and related policy analysis should consider possible impacts on economy society. By coupling model TIMES PanEU impact assessment EcoSense computable general equilibrium NEWAGE, we present an integrated toolbox for European capable internalizing health damage costs air pollution while simultaneously accounting demand changes in services caused by economic feedback loops. effects...
It is widely acknowledged that even as VLSI technology advances, there a looming crisis an important obstacle to the widespread deployment of mobile embedded devices, namely power. This problem can be tackled at many levels like logic, operating systems, micro-architecture and compiler. While have been various proposals for specific compiler optimizations power, has not any attempt systematically map out space possible improvements. In this paper, we quantitatively characterize limits what...
In this paper, we provide a novel compile-time data remapping algorithm that runs in linear time. This is the first fully automatic approach applicable to pointer-intensive dynamic applications. We show can be used significantly reduce energy consumed as well memory size needed meet user-specified performance goal (i.e., execution time) -- relative same application executing without being remapped. These twin advantages afforded by remapped program reduced cache and needs constitute key step...
According to the Paris Agreement, European energy system transition is essential achieve GHG emission reduction targets set by EU and limit growth of global temperatures. Investigating role emerging mitigation technologies their uncertainties together with different crucial realize this transition. This study analyzes electric vehicles' learning paths biomass availability for biofuels, considering policy uncertainties. Since these are considered possible alternatives conventional fuels...
Noise susceptibility and power density have become two limiting factors to CMOS technology scaling. As a solution these challenges, probabilistic (PCMOS) based computing has been proposed. PCMOS devices are inherently that compute correctly with probability p. This paper investigates the trade-offs between energy, performance of correctness (p) inverter. Using simple analytical models delay p inverter, optimum energy product (EDP) value for given constraints is found. The validated using...
In this paper, the results from EU level energy system modelling with TIMES PanEU are interpreted to city development of district heating (DH) systems Warsaw and Helsinki region until 2050. future, there is probably more variation in electricity prices due higher share renewable sources production which further affects operation DH systems. The presented paper indicate that heat storages were almost always profitable investments optimal storage capacity was highest benefit increased prices....
Motivated by the necessity to consider probabilistic approaches future designs, probability and switching energy characteristics of CMOS (PCMOS) circuits are analysed. Using 90 65 nm processes, detailed analytical models for correctness (p) these developed verified through circuit simulations.
In this paper, we provide a novel compile-time data remapping algorithm that runs in linear time. This is the first fully automatic approach applicable to pointer-intensive dynamic applications. We show can be used significantly reduce energy consumed as well memory size needed meet user-specified performance goal (i.e., execution time) -- relative same application executing without being remapped. These twin advantages afforded by remapped program reduced cache and needs constitute key step...