Not available / text

Not available / text

Not available / text

As the feature size of the transistor shrinks into nanometer scale, it becomes a grand challenge for semiconductor manufacturers to achieve good manufacturability of integrated circuits cost-effectively. In this dissertation, we aim at layout optimization algorithms from both manufacturing and design perspectives to address problems in this grand challenge. Our work covers three topics in this research area: a redundant via enhanced maze routing algorithm for yield improvement, a shuttle mask floorplanner, and optimization of post-CMP topography variation. Existing methods for redundant via insertion are all post-layout optimizations that insert redundant vias after detailed routing. In the first part of this dissertation, we propose the first routing algorithm that conducts redundant via insertion during detailed routing. Our routing problem is formulated as a maze routing with redundant via constraints and transformed into a multiple constraint shortest path problem, and then solved by Lagrangian relaxation technique. Experimental results show that our algorithm can find routing solutions with remarkably higher rate of redundant via insertion than conventional maze routing. Shuttle mask is an economical method to share the soaring mask cost by placing different chips on the same mask. Shuttle mask floorplanning is a key step to pack these chips according to certain objectives and constraints related to mask manufacturing and cost. In the second part of this dissertation, we develop a simulated annealing based floorplanner that can optimize these objectives and meet the constraints simultaneously. Chemical-mechanical polishing (CMP) is a crucial manufacturing step to planarize wafer surface. Minimum post-CMP topography variation is preferred to control the defocus in lithography process. In the third of this dissertation, we present several studies on optimization of the variation. First, we enhance the shuttle mask floorplanner to minimize the post-CMP topography variation. Then we study the following singleblock positioning problem: given a shuttle mask floorplan, how to determine a movable block's optimal position with respect to post-CMP topography variation. We propose a fast incremental algorithm achieving 6x to 9x speedup. Finally, we formulate a novel CMP dummy fill problem that targets at minimizing the height variance, which is key to reduce the image distortion by defocus. Experimental results show that with the new formulation, we can significantly reduce the height variance without sacrificing the height spread much.

ITC/USA 2013 Conference Proceedings / The Forty-Ninth Annual International Telemetering Conference and Technical Exhibition / October 21-24, 2013 / Bally's Hotel & Convention Center, Las Vegas, NV / High speed real-time data transportation is most important for telemetry systems, especially for large-scale distributed systems. This paper introduces a STDM (Sync Time Division Multiplexing) network structure for data transportation between devices in telemetry systems. The data in these systems is transported through virtual channels between devices. In addition, a proper frame format is designed based on PCM format to meet the needs of synchronization and real-time transportation in large-scale distributed telemetry systems.

The Arctic is an important component of the Earth’s climate system, and it is a region dynamically coupled to climate phenomena at lower latitudes, through both atmospheric and oceanic paths. The coupling has significant effects on the hydroclimate variability in the Arctic, including effects on sea ice and Arctic precipitation. In this dissertation, we explore the coupling of the lower latitudes and the Arctic hydroclimate through atmospheric mechanisms with dynamical and thermodynamical components, with a focus on the following examples of variability: i) the decadal variability of boreal winter Arctic precipitation, ii) the variability of the strength of the stratospheric polar vortex in boreal winter, and iii) the initial melt of Arctic sea ice in late boreal spring. The goal of the research is to understand what drives the Arctic hydroclimate variability in each of these examples through improved knowledge of the mechanisms linking them to the tropics and Northern Hemisphere midlatitudes. In the first part of the analysis, we explore the mechanisms responsible for the decadal variability of boreal winter Arctic precipitation. We find that the decadal variability of cool-season Arctic precipitation is at least partly connected to decadal modulation of tropical central Pacific sea surface temperatures related to the El Niño-Southern Oscillation (ENSO). The modulation can be described as the oscillation between periods favoring central and eastern Pacific warming events [CPW and EPW, respectively], which are two common types of ENSO variability. By analyzing a collection of CPW and EPW events in reanalysis data, we establish the following connecting mechanism. First, the increase of central Pacific SSTs drive a Rossby wave train that destructively interferes with the zonal wavenumber 1 component of the background extratropical planetary wave in the subpolar region. Next, as a result of this interference, the magnitude of the vertical Rossby wave propagation from the troposphere to the stratosphere decreases and the stratospheric polar vortex strengthens. Finally, the strengthening of the vortex translates into a tendency towards a positive Arctic Oscillation (AO) in the troposphere and a poleward shift of the Northern Hemisphere midlatitude storm tracks, increasing moisture transport from lower latitudes and increasing total Arctic precipitation. In a further investigation of a crucial component of the above mechanism, the initial response of the stratospheric polar vortex to the influence of CPW and EPW is investigated. A 20-member ensemble run of an idealized model experiment in the NCAR Whole Atmosphere Community Climate Model (WACCM) is conducted with prescribed CPW and EPW pattern SST anomalies. Both CPW and EPW events weaken the polar vortex in the ensemble mean. The weakening is mainly tied to changes in the eddy-driven mean meridional circulation, with some contribution from eddy momentum flux convergence. There is a significant spread between ensemble members with identical CPW and EPW forcing, where a few of the ensemble members exhibit a weak strengthening response. The initial conditions of the extratropical atmosphere and subsequent internal variability after the introduction of the CPW and EPW forcing help drive the spread in response between individual members. In the last part of the analysis, using MERRA reanalysis data, the means by which atmospheric eddies affect the trend and variability of the initial melt of Arctic sea ice are explored. We focus specifically on the effects of lower troposphere (i.e. 1000-500 mb average) meridional heat transport by atmospheric eddies, a dynamical component of the atmospheric eddy mechanism, and eddy-generated surface downwelling shortwave and longwave radiation anomalies, a thermodynamical component. Although in a climatological sense, atmospheric eddies in all major frequency bands transport heat poleward into the Arctic, we find that the lower-troposphere eddy meridional heat transport does not contribute to the trend of an earlier initial melt date. However, eddy heat transport still plays an important role in the initialization of individual episodes of initial melt with large areal coverage. In the investigation of two specific episodes, the meridional heat transport term that represents the interaction between the eddy wind and mean temperature fields (i.e. the product of the meridional eddy wind and the mean temperature fields) is most associated with the initial melt in both episodes. Additionally, melt in one of the episodes is also associated with surface downwelling longwave and shortwave radiation anomalies, a result of eddy-generated cloud cover anomalies. Therefore, in individual melt events, the combination of direct eddy meridional heat transport and surface longwave and eddy-driven shortwave radiation anomalies may significantly contribute to the initial melt of Arctic sea ice. This combination may be especially important in episodes where significant initial melt occurs over a large area and over a period of a few days.

The unprecedented and exponential growth of data along with the advent of multi-core processors has triggered a massive paradigm shift from traditional single threaded programming to parallel programming. A number of parallel programming paradigms have thus been proposed and have become pervasive and inseparable from any large production environment. Also with the massive amounts of data available and with the ever increasing business need to process and analyze this data quickly at the minimum cost, there is much more demand for implementing fast data mining algorithms on cheap hardware. This thesis explores a parallel programming model called dataflow, the essence of which is computation organized by the flow of data through a graph of operators. This paradigm exhibits pipeline, horizontal and vertical parallelism and requires only the data of the active operators in memory at any given time allowing it to scale easily to very large datasets. The thesis describes the dataflow implementation of two data mining applications on huge datasets. We first develop an efficient dataflow implementation of a Collaborative Filtering (CF) algorithm based on weighted co-clustering and test its effectiveness on a large and sparse Netflix data. This implementation of the recommender system was able to rapidly train and predict over 100 million ratings within 17 minutes on a commodity multi-core machine. We then describe a dataflow implementation of a non-parametric density based clustering algorithm called Auto-HDS to automatically detect small and dense clusters on a massive astronomy dataset. This implementation was able to discover dense clusters at varying density thresholds and generate a compact cluster hierarchy on 100k points in less than 1.3 hours. We also show its ability to scale to millions of points as we increase the number of available resources. Our experimental results illustrate the ability of this model to “scale” well to massive datasets and its ability to rapidly discover useful patterns in two different applications. / text

published_or_final_version / Electrical Engineering / Master / Master of Philosophy

published_or_final_version / Electrical and Electronic Engineering / Doctoral / Doctor of Philosophy

No description available.