L'intégration tridimensionnelle (3D), où plusieurs puces sont empilées et interconnectées, est en train de révolutionner l'industrie des semi-conducteurs.Cette technologie permet d'associer, dans un même boîtier, des puces électroniques (analogique, numérique, mémoire) avec des puces d'autres domaines(MEMS, bio-capteurs, optique, etc). Cela ouvre de nombreuses voies d'innovation. Néanmoins, l'absence d'outils de conception assistée ordinateur(CAO) adaptés aux systèmes 3D freine l'adoption de la technologie.Cette thèse contribue à deux problématiques liées à la conception 3D : le partitionnement d'un système sur de multiples puces et l'optimisation hiérarchique de systèmes multiphysiques (hétérogènes).La première partie de la thèse est dédiée au problème de partitionner la fonctionnalité d'un système sur de multiples puces. Un outil de « floorplan » 3D a été développé pour optimiser ce partitionnement en fonction de la surface des puces, de la température d'opération du circuit et de la structure des interconnexions. Ce type d'outil étant complexe, nous proposons de régler ses paramètres de façon automatique par l'utilisation d'algorithmes évolutionnaires.Des résultats expérimentaux sur une suite de benchmarks et sur une architecture multi processeur connecté en réseau démontrent l'efficacité et l'applicabilité des techniques d'optimisation proposées.Dans la deuxième partie, nous présentons une méthodologie de conception hiérarchique qui est adaptée aux systèmes hétérogènes. La méthode combine une approche ascendante et descendante et utilise des courbes de compromis(Fronts de Pareto) comme une abstraction de la performance d'un circuit.La contribution principale de la thèse consiste à utiliser des techniques d'interpolation pour représenter les Fronts de Pareto par des fonctions continues et à leur intégration dans des processus d'optimisation classiques. Cela permet un gain en flexibilité lors de l'étape ascendante du flot (caractérisation) et un gain en temps lors de l'étape descendante (synthèse). Le flot de conception est démontré sur un amplificateur opérationnel ainsi comme sur la synthèse d'un lien optoélectronique avec trois niveaux hiérarchiques. / 3D integration technology is driving a strong paradigm shift in the design of electronic systems. The ability to tightly integrate functions from different technology nodes (analog, digital, memory) and physical domains (MEMS, optics, etc) offers great opportunities for innovation (More than Moore). However, leveraging this potential requires efficient CAD tools to compare architectural choices at early design stages and to co-optimize multiphysics systems.This thesis work is divided into two parts. The first part is dedicated to the problem of partitioning a system into multiple dies. A 3D floorplanning tool was developed to optimize area, temperature and the interconnect structure of a 3DIC. Moreover, a meta-optimization approach based on genetic algorithms is proposed to automatically configure the key parameters of the floorplanner. Tests were carried out on architectural benchmarks and a NoC based multiprocessor to demonstrate the efficiency of the proposed techniques.In the second part of the thesis, a hierarchical design methodology adapted to heterogeneous systems is presented. The method combines the bottom-up and top-down approaches with Pareto-front techniques and response surface modeling. The Pareto front of lower level blocks are extracted and converted into predictive performance models that can be stored and reused in a top-down optimization process. The design flow is demonstrated on an operational amplifier as well as on the synthesis of an optoelectronic data link with three abstraction levels.

No description available.

The process of designing a floorplan is highly iterative and requires extensive human labor. Currently, there are a number of computer programs that aid humans in floorplan design. These programs, however, are limited in their inability to fully automate the creative process. Such automation would allow a professional to quickly generate many possible floorplan solutions, greatly expediting the process. However, automating this creative process is very difficult because of the many implicit and explicit rules a model must learn in order create viable floorplans. In this paper, we propose a method of floorplan generation using two machine learning models: a sequential model that generates rooms within the floorplan, and a graph-based model that finds adjacencies between generated rooms. Each of these models can be altered such that they are each capable of producing a floorplan independently; however, we find that the combination of these models outperforms each of its pieces, as well as a statistic-based approach.

År 2020 drabbades hela världen av en pandemi. För att försöka hindra den snabba smittspridningen har Folkhälsomyndigheten sedan mars 2020 rekommenderat arbetsgivare att låta deras arbetsgivare arbeta hemifrån. Samtidigt som vi numera rekommenderas att arbeta hemifrån har bostäderna det senaste åren bara blivit mindre och mindre. Studier visar på att vi kommer vilja fortsätta arbeta hemifrån även efter pandemin, därför tros pandemin bli en vändande punkt för bostadsbyggandet.  Denna studie har som syfte att undersöka om den plötsliga efterfrågan på hemarbete kan komma att påverka bostäders utformning i framtiden samt hur en arbetsplats kan integreras i eller i samband med en bostad. För att besvara syftet har en litteraturstudie, fallstudie samt egen konceptuell utformning gjorts.  Litteraturstudien gjordes för att undersöka studier gjorda angående hemarbete under Covid-19-pandemin samt äldre studier om hemarbete generellt. I litteraturstudien undersöktes även de svenska byggregler som ligger till grund för studierna i fallstudien och planlösningarna som presenteras i resultaten.  I fallstudien undersöks sju olika lägenheter indelade i fyra kategorier: kontorsplats i ett rum i lägenhet, arbetsrum i lägenhet, avskilt kontor i samband med bostad, samt gemensam plats för kontorsplatser i flerbostadshus. Dessa kategorier används sedan även för den konceptuella utformningen.  Den konceptuella utformningen baserades på litteratur- och fallstudien. För att verklighetsförankra resultatet valdes två av de studerade fallen i fallstudien som mall för utformningen. Resultatet blev ett antal olika planlösningar indelade i samma kategorier som fallstudien.  Slutsatsen som kan dras från detta arbete är att det är mycket troligt att Covid-19- pandemin och hemarbete kommer att behöva ändra hur vi bygger bostäder idag. Denna studie visar på ett antal olika sätt som hemarbete kan integreras i eller i samband med en bostad. / In the year 2020, the world was hit by a pandemic. In an attempt to slow down the rapid spread of the infection the Swedish Public Health Agency has since March 2020 recommended employers to let their employees work from home. In the same time as we are recommended to work from home, housing has in recent years only become smaller and smaller. Studies show that we will want to continue working from home even after the pandemic, therefore the pandemic is believed to be a turning point for housing construction.  The purpose of this study is to investigate whether the sudden demand for working from home may affect the design of residences in the future and how a workplace can be integrated in or in correlation to a residence. To answer the purpose, a literature study, case study and own conceptual design have been done.  The literature study was made by examining published studies about working from home during the Covid-19-pandemic and older studies on working from home in general. The literature study also examined the Swedish building regulations that form the basis for the studies in the case study and the floor plans presented in the results.  The case study examines seven different apartments divided into four categories: office space in one room in an apartment, work space in an apartment, separate office in connection with housing, and common space for office space in apartment buildings. These categories are then also used for the conceptual designing.  The conceptual designing was based on the literature and the case study. In order to make the results true too real life, two of the cases studied were chosen in the case study as a template for the designing. The conceptual designing resulted in a number of different floor plans divided into the same categories as the case study.  The conclusion of this study is that it is very likely that the Covid-19 pandemic and working from home will change how we build residences today. This study shows a number of different ways in which working from home can be integrated into or in correlation to a residence.

Abstract Purpose: Important for house manufacturers is to keep track of specific customer preferences and priorities to have products that matches these. If a manufacturer does not have a product that fits the customer priorities, manufacturers will lose customers both before and during the process has started. The goal of the study is to report what house manufacturers should prioritize in the floor plan design of prefabricated two-storey houses. Method: The study has been conducted with a variety of methods. In order to study customer preferences and priorities a survey was made for potential house byers and interviews with house manufacturers sales agents. To get an overview about the changes that are most commonly performed, an analysis of documents on two of Götenehus house models was made. Findings: The results of the study show the kitchen's high priority among potential house buyers. Clearly shows that, above all, the social spaces in the form of kitchen living and dining room is generally very high priority among the general population In Stockholm. The study also shows that these rooms were desired with a relatively large surface area. It was also very unusual for these rooms to get reduced space. Unlike the social spaces, the private spaces hade ha wide range of demands. Above all, the bathroom and the master bedroom had highly prioritized. The other bedrooms were however generally one of house buy’ers lowest priority. In general, the service areas were shown to be the lowest prioritised rooms. The laundry room was desired to be large with great storage units, but were rarely executed as wanted. Implications: The social space, master bedroom and the bathroom has a great importance for customers in Stockholm County. House manufacturers should prioritize these rooms. In contrast, house manufacturers should not prioritize room as office and WC/D. Doing so, house manufacturers can offer customers a product that minimizes the impact on costs by changes. Limitations: The study deals only with prefabricated detached two-storey houses. The study does not include factors such as finance, property space or family constellations. The study was limited to the Stockholm County because of the concentrated market for two-storey house. The gathered information was not extracted prior ten years. / Sammanfattning Syfte: Det är viktigt för hustillverkarna att deras produkteker är anpassade efter hur kundernas önskemål ser ut och hur dessa prioriteras. Ifall produkterna inte uppfyller kundernas värderingar kan det leda till att företaget både tappar kunder innan och under köpprocessen. Målet med studien är därför att redovisa vilka rum hustillverkare bör prioritera vid planlösningsutformning av prefabricerade tvåplanshus. Metod: Studien har utförts med ett flertal metoder. För att kunna studera kundernas önskemål och prioriteringar, utfördes en enkätundersökning mot potentiella husköpare, samt intervjuer med Götenehus säljagenter i Stockholm. För att få en överblick kring vilka ändringar som är mest förekommande utfördes en dokumentanalys på två av Götenehus husmodeller. Resultat: Studiens resultat visar att potentiella husköpare prioriterar de sociala utrymmena kök och vardagsrum högst i bostaden. Kunderna har stora önskemål om stora och öppna ytor. Generellt i de privata utrymmena är det önskemål om stora ytor. Däremot prioriterades de privata rummen olika av kunden. Framför allt har badrummet och föräldrasovrummet en hög prioritet medan de övriga sovrummen tillsammans med allrum-met och arbetsrummet prioriterades betydligt lägre. Överlag var serviceutrymmena lågt prioriterade av kunderna. Det fanns önskemål om väl tilltagna ytor i tvättutrymmet men det var inget kunderna var beredda att förändra. Förvaringsmässigt var det endast i föräldrasovrummet som kunderna önskade en klädkammare. I resterande rum som önskan om förvaring fanns, räckte det med en garderobslösning. Konsekvenser: Bostadens sociala utrymmen samt föräldrasovrummet och badrummet har en stor betydelse för kunderna i Stockholms län. Dessa rum bör hustillverkare prioritera vid utformning av planlösningen. Däremot bör hustillverkarna inte prioritera rum som arbetsrummet och WC/D vid utformningen. Görs detta kan hustillverkarna erbjuda kunderna en produkt som minimerar kostnadspåverkande ändringar. Begränsningar: Studien behandlar endast prefabricerade fristående villor i två plan. Studien omfattar inte faktorer som påverkar hur en planlösning utformas som exempelvis ekonomi, tomtyta och familjekonstellationer. Studien begränsades till Stockholms län eftersom denna marknad är stor för tvåplanshus. Studien är tidsavgränsning till 2006 vilket är en 10 årsperiod.

Capacitive crosstalk noise can affect the delay of a switching signal or induce a glitch on a static signal causing timing violations or chip failure. Crosstalk noise depends on coupling parasitics, driver strength, signal timing characteristics, and signal transition patterns. Layout level crosstalk analysis techniques are generally pessimistic and computationally expensive for large designs due to lack of design flexibility at lower-levels of design hierarchy. The architectural decisions such as type of interconnect architecture, number of storage and execution units, network of communicating units, data bus width, etc., have a major impact on the quality of design attributes such as area, speed, power, and noise. To address all these concerns, we propose a high-level synthesis framework to optimize for worst-case crosstalk patterns on coupled nets, a floorplan driven high-level synthesis framework to minimize coupling capacitance, and an on-chip technique to dynamically detect and eliminate worst-case crosstalk pattern on bus-based macro-cell designs. Due to Miller coupling effect, the switching activity pattern on adjacent nets may increase the effective capacitance seen by a victim net and thereby it may cause a worst-case signal delay on the victim net. However, signal activity pattern on coupled nets are dependent on data correlations which in turn depend on resource sharing. The resource sharing in turn depends on scheduling, allocation, and binding during high-level synthesis flow. Therefore, we propose a Simulated Annealing (SA) based design space exploration of HLS design subspace, bus line re-ordering, and encoding subspaces to optimize for worst-case crosstalk pattern in bus-based macro-cell designs. We demonstrate that the proposed framework will aid layout level techniques in eliminating false positive violations. We also propose an SA based algorithm to explore floorplan and HLS subspaces to optimize coupling capacitances in bus-based macro-cell designs. We have integrated an RTL floorplanner in HLS flow to estimate coupling capacitances between bus lines. Crosstalk analysis using Cadence Celtic shows that the designs generated by the proposed framework results in less number of crosstalk violations compared to designs generated through traditional ASIC design flow. We also propose an on-chip crosstalk detection and elimination technique that dynamically detects and eliminates worst-case crosstalk pattern with minimum area penalty compared to other layout level techniques reported in the literature.

I dagsläget har samhällsutvecklingen en hög takt vilket innebär ständiga förändringar i omvärlden. Bostäder är något som bland annat förändras i takt med samhällsutvecklingen. I och med utvecklingen behövs planlösningar som är enkla att förändra. För att åstadkomma detta kan projektörer beakta framtida möjligheter för förändringar i planlösningar under projekteringsskedet. I arbetet har flexibilitet i planlösningar hos bostäder studerats. Flexibilitet innebär kortfattat att förändringar i planlösningar ska kunna göras enkelt och utan någon större kostnad. Fokus har varit på planlösningar i bostäder från två tidsperioder, därav miljonprogrammet (Bostadsbyggande i Sverige år 1965-1975) och nutid (år 2005-2015). Arbetet beskriver om typen av planlösning påverkar flexibiliteten, om flexibilitet är förändrat över tiden, varför flexibilitet i bostäder behövs och vad god flexibilitet i planlösningar ger möjlighet till. Lagstiftning om bostadsutformning har studerats för att förstå vad som styrt bostadsutformningen under respektive tidsperiod. En viktig skillnad mellan lagstiftningarna är att man gått från normkrav till funktionskrav. Det har även studerats om människans boendeförhållanden i bostaden från 1960-talet och framåt för att förstå bostadsutvecklingen. Arbetet är baserat på en kvalitativ metod där fakta främst insamlats från publicerade böcker som lästs textkritiskt. En analys av planlösningar från miljonprogrammet och nutiden har gjorts. Metoden blev framgångsrik genom att bra kriterier gjordes innan analysen som kunde indikera på om planlösningarna hade god flexibilitet eller inte. Analysen av planlösningarna från respektive period gjordes för att undersöka om flexibilitet är förändrat över tiden. I analysen jämfördes öppna och slutna planlösningar för att studera om planlösningstypen påverkar flexibiliteten, resultatet av arbetet tyder på att planlösningstypen påverkar flexibiliteten. Slutna planlösningar representerar miljonprogrammets tid och öppna planlösningar representerar nutid. / The development of Swedish society today is happening fast. With this comes that the surrounding enviroment of residental area is facing constant changes. The recidences change with the same speed as the social development. As a result floorplans of recidences have to be easy to change. To achieve easy transformable floorplans planners should pay attention to the possibility of transforming during the design stage. In this work flexible floorplans of residences have been studied. Flexible floorplans mean that changes can be made easily to the floorplan and at a low cost. The focus of this work have been to look at residential floorplans from two different periods. The first period is the time of the "million programme" (residential building programme during year 1965-1975 in Sweden) and the second period is the present time (year 2005-2015). The work describes why flexible floorplans are needed, which opportunities the flexibility gives, if flexibility is time-related or unchanged over time and if an open floorplan or closed floorplan affects the flexibility. Legislation regarding residential design has been studied to understand which rules have controlled the floorplannings during the different times. One essential difference between legislation and rules from the time periods is that in the first period, the law set the standard requirements and nowadays it sets functional requirements. The human way of life in the housing has been studied from the 1960´s until today to understand the requirements of housing development. The work is based on a qualitative method where facts mainly have been collected from published books that have been read critically. An analysis of floorplans from the "millione programme" and the present has been performed. The method has been successful due to having clear criteria set before the analysis to indicate if the floorplan had good flexibility or not. The analysis of the floorplans was made for each time period to examine if flexibility is time-related or unchanged over time. In the analysis open and closed floorplans were compared to see if the type of plan affects the flexibility. Closed floorplans represent the "million programme" period and open floorplans represent the present.

Parsing floorplans have been a problem in automatic document analysis for long and have up until recent years been approached with algorithmic methods. With the rise of convolutional neural networks (CNN), this problem too has seen an upswing in performance. In this thesis the task is to recover, as accurately as possible, spatial and geometric information from floorplans. This project builds around instance segmentation models like Cascade Mask R-CNN to extract the bulk of information from a floorplan image. To complement the segmentation, a new style of using keypoint-CNN is presented to find precise locations of corners. These are then combined in a post-processing step to give the resulting segmentation. The resulting segmentation scores exceed the current baseline of the CubiCasa5k floorplan dataset with a mean IoU of 72.7% compared to 57.5%. Further, the mean IoU for individual classes is also improved for almost every class. It is also shown that Cascade Mask R-CNN is better suited than Mask R-CNN for this task.

Ett stenkast från stadsdelen Liljeholmen i Stockholm ligger området Lövholmen. Ett idag till stor del övergivet industriområde där man möts av krossade rutor, förfallna byggnader och nedklottrade fasader. Stockholms stad har tillsammans med en rad olika byggherrar gjort en stadsplanering som är ett förslag om ändring i detaljplan, från ett industriområde till ett bostadsområde. Stadsplaneringen Strukturplan och ramverk för Lövholmen innehåller föreslagna riktlinjer om utformningen för en ny stadsdel. Syftet är att utforma ett gestaltningsförslag till en av fastigheterna enligt den nya detaljplanen som om den vunnit laga kraft, me även att undersöka hur planlösningar kan påverka människans hälsa och trivsel. Förslaget kommer förhålla sig till strukturplanens direktiv där målet är att göra Lövholmen till en tryggare och trevligare plats. Detta kommer göras med hjälp av analyser och undersökningar från allmänhetens önskemål och synpunkter. Resultatet, Arken, kommer visa ett bostadsförslag med en arkitektonisk anknytning till platsen som kommer göra Lövholmen till en mer harmonisk plats. Bostadsförslaget kommer medföra gemenskap för människor som bosätter sig i kvarteret samt genom genomarbetade planlösningar som grundar sig i undersökningar kring människans välmående. / A few steps away from the district of Liljeholmen in Stockholm lies the area of Lövholmen. A largely abandoned industrial zone where visitors are greeted by crushed window panes, dilapidated buildings and defaced facades. Stockholm City has along with a number of different developers made a proposal for a change in the detailed development plan, from an industrial area to a residential area. The urban planning Strukturplan och Ramverk för Lövholmen contains proposed guidelines for the design of a new neighborhood. Our purpose in this thesis is to design a residential proposal as if the detailed development plan has won legal force, also to examine how floor plans can affect human health and well-being. The proposal will relate to urban development guidelines by Stockholm stad, where the aim is to make Lövholmen a safer and nicer place. This will be done with the help of analyzes and surveys from the public's wishes and views. The result of our vision; Arken, will consist a housing proposal with an architectural connection to the place that will make Lövholmen a more harmonious site. The housing proposal will bring a sense of community to people who settle in the neighborhood as well as elaborated floor plans based on surveys on human well-being.

This master thesis introduces FGSSNet, a novel multi-headed feature-guided semantic segmentation (FGSS) architecture designed to improve the generalization ability of segmentation models on floorplans by injecting domain-specific information into the latent space, guiding the segmentation process. FGSSNet features a U-Net segmentation backbone with a jointly trained reconstruction head attached to the U-Net decoder, tasked with reconstructing the injected feature maps, forcing their utilization throughout the decoding process. A multi-headed dedicated feature extractor is used to extract the domain-specific feature maps used by the FGSSNet while also predicting the wall width used for our novel dynamic scaling algorithm, designed to ensure spatial consistency between the training and real-world floorplans. The results show that the reconstruction head proved redundant, diverting the networks attention away from the segmentation task, ultimately hindering its performance. Instead, the ablated reconstruction head model, FGSSNet-NoRec, showed increased performance by utilizing the injected features freely, showcasing their importance. FGSSNet-NoRec slightly improves the IoU performance of comparable U-Net models by achieving 79.3 wall IoU(%) on a preprocessed CubiCasa5K dataset while showing an average IoU increase of 3.0 (5.3%) units on the more challenging real-world floorplans, displaying a superior generalization performance by leveraging the injected domain-specific information.