In the last decades, the main driving force behind the astonishing development of CMOS technology, was the transistor scaling process. The reduction of transistor sizes has granted a continuous boost in circuits performance. But now that the scaling process is reaching its physical limits, researchers are forcusing on new emerging technologies. Research on these new technologies is usually carried on using a traditional approach. Some studies concentrate on new devices without analyzing circuits based on them. Other studies analyze circuit architectures without considering devices characteristics and limitations. However, given that the nature of emerging technologies can be very different from CMOS, new research methodologies should be adopted. A clear link between device and architectural analysis is necessary to understand the true potential of the technology under study. The objective of this PhD thesis is the analysis of emerging technologies using an innovative methodology. Using complex and realistic circuits as benchmark, high level models are built incorporating low level device characteristics. This methodology strongly links device and architectural levels. The methodology was applied to two emerging technologies: NanoMagnet Logic (NML) and Nanoscale Application Specific Integrated Circuits (NASIC). A brief introduction of fundamental information on the two technologies is given in Chapter 1. The application of the methodology on NML technology is divided in two parts (Chapter 2): i) architecture-level timing and performance analysis and circuits optimization; (ii) area and power estimations using VHDL modeling. Starting from an exhaustive analysis of the effects and the consequences derived by the presence of loops in a complex NML sequential architecture, solutions have been proposed to address the problem of signal synchronization, and optimization techniques have been explored for performance maximization. Area and power estimations have been performed on multiple NML architectures in order to obtain a complete evaluation on the implementation of NanoMagnet Logic in comparison with the CMOS technology. Chapter 4 is dedicated to NASIC technology with basic principles described in Chapter 3. Basic computational blocks are implemented using a multilevel modeling approach. A detailed analysis of circuits' area and power estimations is obtained. Techniques to optimize the area of circuits at the cost of reduced throughput were also investigated. The research activity presented in this thesis highlights the development of an innovative methodology based on high-level models that embed information obtained from physical level simulations. By exploiting this methodology to different emerging technologies, such as NML and NASIC, it allows to eciently analyze circuits and therefore to bring architectural improvements.
http://hdl.handle.net/11583/2644810
Titolo: | Multilevel Modeling and Architectural Solutions for Emerging Technology Circuits |
Autori: | |
Data di pubblicazione: | 2016 |
Abstract: | In the last decades, the main driving force behind the astonishing development of CMOS technology..., was the transistor scaling process. The reduction of transistor sizes has granted a continuous boost in circuits performance. But now that the scaling process is reaching its physical limits, researchers are forcusing on new emerging technologies. Research on these new technologies is usually carried on using a traditional approach. Some studies concentrate on new devices without analyzing circuits based on them. Other studies analyze circuit architectures without considering devices characteristics and limitations. However, given that the nature of emerging technologies can be very different from CMOS, new research methodologies should be adopted. A clear link between device and architectural analysis is necessary to understand the true potential of the technology under study. The objective of this PhD thesis is the analysis of emerging technologies using an innovative methodology. Using complex and realistic circuits as benchmark, high level models are built incorporating low level device characteristics. This methodology strongly links device and architectural levels. The methodology was applied to two emerging technologies: NanoMagnet Logic (NML) and Nanoscale Application Specific Integrated Circuits (NASIC). A brief introduction of fundamental information on the two technologies is given in Chapter 1. The application of the methodology on NML technology is divided in two parts (Chapter 2): i) architecture-level timing and performance analysis and circuits optimization; (ii) area and power estimations using VHDL modeling. Starting from an exhaustive analysis of the effects and the consequences derived by the presence of loops in a complex NML sequential architecture, solutions have been proposed to address the problem of signal synchronization, and optimization techniques have been explored for performance maximization. Area and power estimations have been performed on multiple NML architectures in order to obtain a complete evaluation on the implementation of NanoMagnet Logic in comparison with the CMOS technology. Chapter 4 is dedicated to NASIC technology with basic principles described in Chapter 3. Basic computational blocks are implemented using a multilevel modeling approach. A detailed analysis of circuits' area and power estimations is obtained. Techniques to optimize the area of circuits at the cost of reduced throughput were also investigated. The research activity presented in this thesis highlights the development of an innovative methodology based on high-level models that embed information obtained from physical level simulations. By exploiting this methodology to different emerging technologies, such as NML and NASIC, it allows to eciently analyze circuits and therefore to bring architectural improvements. |
Appare nelle tipologie: | 8.1 Doctoral thesis Polito |
File in questo prodotto:
File | Descrizione | Tipologia | Licenza | |
---|---|---|---|---|
PhDThesis_Wang Juanchi.pdf | Doctoral Thesis | 5. Doctoral Thesis | PUBBLICO - Licenza CC by-nc-nd | Visibile a tuttiVisualizza/Apri |