4987 research outputs found
Sort by
Cryptonight Gpu Mining Efficiency
Get PDFThe purpose of this thesis is to study the efficiency of using graphical processing units in Cryptonight, the proof-of-work system used to mine Monero. By understanding the dependence of Cryptonight in memory, we theorize that by improving read and write delays we can improve mining results.
In this thesis, there is a major focus on the technology behind Bitcoin and Monero since at the time of writing stand to be the most respectable ecosystems. The paper starts by analyzing the history of proof of work and how it has evolved during the past few years. We study the use of CPUs and GPUs to mine during the lifetime of Bitcoin and the eventual development of specialized ASICs. How GPU mining is the current best solution for mining Monero because of its commitment to stay ASIC resistant and why GPU mining is the best way to build a general-purpose miner that has the flexibility to mine different coins and different algorithms.
We look at all the hardware components required to build a GPU miner, how to choose between alternatives and how this affects efficiency. During this writing and testing period many components were burned or damaged so some of the common mistakes in handling hardware will be mentioned. We will take a look at all the hardware modifications that can be made like overclocking, undervolting and modifying bios memory timings to increase mining efficiency measured in hash/watt units.
Major focus is put in understanding memory timings, how changing specific values impacts hashrate, measuring this data to quantify the efficiency benefits that can be used in profitable mining.
This thesis is an attempt to document as much as possible of the knowledge that has been flowing around lately as interest on crypto currencies has increased in the past few years
Smile Recognition Implementation on Embedded Platforms
Get PDFIn this work, our focus is on the real-time development of a smile recognition system on low resource computational devices utilizing deep learning algorithms which could be simply further developed to address issues in mentioned areas.
We have primarily used the Looking at People (LAP) dataset for training and testing various neural network architectures. Images in this dataset have been pre-processed at first by acts of cropping around the facial area and face alignment. Then six pre-trained deep learning network architectures were finetuned for this purpose.
The fine-tuned models were deployed on Nvidia’s embedded platform and we were employing an asynchronous design to provide smoother frame rate through parallelization and multithreading. Accuracy and speed of these models were retrieved letting us compare them to each other and choose the most suitable ones for this task. Our research shows that modern low complexity architectures could almost reach the older or bulkier ones’ performance
Developing a product policy to overcome the challenges of a company’s sales process
Get PDFDiplomityön kohdeyrityksenä oli suomalainen tuotantojärjestelmiä valmistava yritys. Yrityksen tuotantojärjestelmät automatisoivat asiakkaiden tuotantoprosesseja. Kohdeyritys on perinteinen projektitalo, joiden tuotteissa on kuitenkin myös vakioituja rakenteita. Yritys toimii kansainvälisesti kilpaillulla alalla. Yrityksessä oli havaittu lukuisia tuotepolitiikan puutteeseen liittyviä ongelmia. Tuotepolitiikan luominen olikin kirjattu yrityksen strategisiin tavoitteisiin.
Diplomityön tavoitteena oli kehittää tuotepolitiikka yrityksen myyntiprosessin haasteiden ratkaisemiseksi. Tätä lähdettiin tavoittelemaan kolmen eri vaiheen, kirjallisuuskatsauksen, nykytilan kartoituksen ja benchmarking-haastattelujen kautta. Kirjallisuuskatsauksella selvitettiin, mitä tarkoittavat tuote ja tuotepolitiikka. Tuotepolitiikan havaittiin olevan työkalu, jolla kommunikoidaan tuotestrategia eteenpäin. Se lukee alleen muun muassa tuotteen ja tuotestrategiat. Sen avulla yritys kertoo, missä strategisessa asemassa se pyrkii olemaan tuotteellaan. Kun tämä on tiedossa, yritys voi keskittyä ydinosaamiseensa.
Nykytilan kartoitus suoritettiin puolistrukturoituina teemahaastatteluina yrityksen sisällä. Nykytilan kartoituksessa selvitettiin, mikä on tuotepolitiikan nykytilanne kohdeyrityksessä. Tuotepolitiikan nykyinen taso nähtiin puutteellisena. Yrityksen jokaisessa toiminnossa nähtiin tähän liittyviä ongelmia. Ongelmat liittyivät pääasiassa siihen, ettei yrityksen sisällä ollut yksimielisyyttä siitä, mikä on yrityksen ydinosaamista. Ongelmiin haettiin ratkaisuja benchmarking-haastattelujen kautta, jotka suoritettiin puolistrukturoituina teemahaastatteluina neljässä eri yrityksessä.
Muilla yrityksillä oli lukuisia eri toimintatapoja ylittää kohdeyrityksessä esiintyvät ongelmat. Näitä esille tulleita toimintatapoja hyväksikäyttäen ja soveltamalla teoriaosuutta, luotiin kohdeyritykselle tuotepolitiikan dokumentti. Tuotepolitiikan dokumentti on myynnin työkalu, jolla myyntiprosessin ongelmat voidaan ylittää. Dokumentti auttaa yritystä myös keskittymään ydinosaamiseensa selkeyttämällä tuotteiden tavoitteita ja strategioita. Luodulla tuotepolitiikan dokumentilla kommunikoidaan tuotepolitiikka yrityksen sisällä. Kun yrityksessä täytetään kyseinen dokumentti, yrityksessä otetaan huomioon tuotteen teknisten ominaisuuksien lisäksi muun muassa tuotteen toimintaympäristö
Python-ohjelmien suorituskyvyn parantaminen
Get PDFPython on moderni ohjelmointikieli, jolla ohjelmien kehitys on monia muita kieliä nopeampaa, mutta se ei yksinään sovi useimpiin suorituskykykriittisiin käyttökohteisiin. Tämän suorituskykyongelman ratkaisemiseksi on luotu useita ratkaisuja, jotka eroavat toisistaan lähestymistavassaan. Ratkaisut eroavat myös suuresti siinä kuinka paljon alkuperäistä Python-ohjelmaa tulee muokata suorituskykyongelman ratkaisemiseksi.
Tämän työn tarkoitus on vertailla esimerkkiohjelman avulla kolmea eri tapaa, jolla Python-ohjelmien suorituskykyä voidaan parantaa. Esimerkkiohjelmasta tehtyjen toteutuksien suorituskykyä vertaillaan keskenään. Samalla saavutetaan jonkinasteinen käsitys siitä, kuinka paljon alkuperäistä ohjelmaa tulee muokata, jotta voidaan hyödyntää eriratkaisutapoja.
Kaikilla työssä vertailluilla ratkaisuilla saavutettiin merkittävä parannus testiohjelman suorituskykyyn. Cythonilla ja Shed Skinillä saavutettiin parhaat tulokset. PyPyn tuoma suorituskykyparannus hieman heikompi, mutta sen hyödyntämistä varten testiohjelmaan ei tarvinnut tehdä lainkaan muutoksia. Cythonia varten testiohjelmaa tuli muokata eniten, sillä toteutus hyödynsi C++:n tietorakenteita
Implementation of Linux drivers as kernel modules and in user mode
No full textLinux-ajureita voidaan valmistaa ja suorittaa sekä ydintilassa että käyttäjätilassa. Ongelmana on, ettei välttämättä tiedetä kumpi tapa olisi parempi ratkaisu laiteajurin valmistukseen. Työssä toteutettiin vertailu ydintilan ja käyttäjätilan ajurien tärkeimpien eroavaisuuksien välillä, ja analysoitiin niiden vaikutusta eri laiteajurityyppien valmistukseen. Havaittiin, että käyttäjätilan ajurit voivat olla rajoittuneempia, mutta moniin ongelmiin on kiertokeino. Kummallekin ajurityypille löytyi tapauksia, joissa ne ovat toista parempia
Sekoittumattoman polymeeriseoksen sekoittuvuuden parantaminen grafeenioksidin avulla
Get PDFIn this thesis, normally highly immiscible blends of low-density polyethylene and polyamide 66 were created and studied. Blends, which are based on engineering thermoplastic, like polyamide, and on commodity plastic, like polyethylene, are an interesting research target because it would be possible to create blends, with many desired properties and low cost. These blends could be then exploited in many applications in different fields. The usual problem of polymer blending is related to the different chemical structures of blend’s components. This leads to phase separation in the blend, which leads to poor properties. To tackle this problem, a third component can be added to the blend. These components are called compatibilizers and they are usually block or graft copolymers but using them in blends has many challenges. That is why in this work, graphene oxide was used as a compatibilizer to research its potential to enhance blends properties.
The theoretical part of the thesis presents information about low-density polyethylene, polyamide66, polymer blending and graphene. The basic properties, structural information and typical uses of low-density polyethylene and polyamide 66 are discussed first. Followed by this, is a section of polymer blending. Polymer blending is introduced by providing information about extruders, mixing processes and thermodynamics. Also, compatibilizers are introduced. Finally, graphene is presented. The focus is mainly on graphene oxide and reduced graphene oxide, providing information about their structures, properties and ways to produce them. The functionalization of graphene oxide is also presented.
The experimental part included modified Hummers’ method to produce graphene oxide and preparation of suitable masterbatches for processing. Melt compounding with twin-screw extruder was performed to mix the blends and to potentially partially reduce and functionalize the graphene oxide. Various characterization methods were used in this work with suitable programs to analyze the samples and to see if there were any indications about successful blending.
Several characterization methods indicate, that the dispersion of polyamide 66 in low-density polyethylene matrix was enhanced with the use of partially reduced and non-covalently functionalized graphene oxide as a compatibilizer. This is a key thing in blending and in enhancing blends properties. Also, positive results on enhanced thermal properties and chemical interactions between components are witnessed
Fixed Switching Frequency Direct Model Predictive Control for Grid-Connected Converters
No full textModel predictive control (MPC) has recently been gaining ground as a suitable control method for power electronic converters. It is formulated as an optimization problem in the time domain subject to certain constraints. Moreover, it is able to handle multiple-input multiple-output (MIMO) switched (non)linear systems and implement limitations in the form of hard and/or soft constraints.
The important challenges in controlling grid-connected converters involve ensuring that the output harmonic spectra comply with specific grid codes and that fast responses are achieved during changes in power references. In grid-connected applications, most often the produced harmonic spectra need to be well-defined to meet the grid codes. A fixed switching frequency and a symmetrical switching pattern ensure discrete harmonic spectra, making the compliance with the grid codes easier. In addition, fast responses during transients can be achieved by eliminating the modulator, i.e., direct control.
This thesis presents a direct MPC algorithm for a three-phase two-level grid-connected voltage source converter (VSC) with an LCL filter that can operate the converter at a fixed switching frequency despite the absence of a modulator. The performance of the proposed method is compared to open-loop carrier-based pulse width modulation (CB-PWM) and the IEEE519 grid code. Several refinements to the algorithm are presented which improve the performance of the system. Moreover, the algorithm is extended to emulate the 120 degree discontinuous PWM switching pattern. In steady-state operation, the method achieves similar total harmonic distortion (THD) levels to CB-PWM, and during transients faster responses are obtained due to the elimination of the modulation stage
Puhelinkoppien koostumuksien vaikutukset terveyteen ja koostumuksen määrittelyprosessin kehittäminen
Get PDFOpen-plan offices have become popular in the recent decades and along with the popularity, noise has become a general problem in the offices around the world. Framery Oy, the case company of this thesis, manufactures phone booths that are being offered as a solution to the noise problem in open-plan offices. The case company has faced demands from the customers to have knowledge about what materials and substances the phone booths are composed of. To fulfill this demand the company has recognized that it needs Health Product Declaration (HPD) documents for their products. In practice these documents declare what materials and substances compose the phone booths.
This thesis studies what are the materials that compose the phone booths, how the composition determination process could be developed, do some of the materials cause hazards for the health of the end user and is there any alternatives for these hazardous materials. As a result the composition of the booths was recognized, a generalized process model was developed to help define the material composition of a product, two different possibly hazardous materials were recognized and few less hazardous alternatives were found for these two different materials
Autonomous Navigation of Mobile Robots: Marker-based Localization System and On-line Path
Get PDFTraditional wheelchairs are controlled mainly by joystick, which is not suitable solution with major disabilities. Current thesis aiming to create a human-machine interface and create a software, which performs indoor autonomous navigation of the commercial wheelchair RoboEye, developed at the Measurements Instrumentations Robotic Laboratory at the University of Trento in collaboration with Robosense and Xtrensa,. RoboEye is an intelligent wheelchair that aims to support people by providing independence and autonomy of movement, affected by serious mobility problems from impairing pathologies (for example ALS – amyotrophic lateral sclerosis).
This thesis is divided into two main parts – human machine interface creation plus integration of existing services into developed solution, and performing possible solution how given wheelchair can navigate manually utilizing eye-tracking technologies, TOF cameras, odometric localization and Aruco markers.
Developed interface supports manual, semi-autonomous and autonomous navigation. In addition to that following user experience specific for eye-tracking devices and people with major disabilities. Application delevoped on Unity 3D software using C# script following state-machine approach with multiple scenes and components.
In the current master thesis, suggested solution satisfies user’s need to navigate hands-free, as less tiring as possible. Moreover, user can choose the destination point from defined in advance points of interests and reach it with no further input needed. User interface is intuitive and clear for experienced and inexperienced users. The user can choose UI’s icons image, scale and font size. Software performs in a state machine module, which is tested among users using test cases. Path planning routine is solved using Dijkstra approach and proved to be efficient
Verkkosivun selailutietojen käsittely ja visualisointi kohdistetun sisällön ratkaisuja varten
Get PDFDue to the rapidly increasing amounts of information on the Internet, providing users with relevant content becomes increasingly important. To achieve this, websites may employ recommender systems to offer targeted content to their users. In addition, due to the amount of information, recommender systems have begun adapting machine learning algorithms to more efficiently and accurately provide relevant content for the user.
Machine learning based approaches to recommender systems often involve ”teaching” the system by feeding them pre-existing data about users and their preferences. This study aims to provide a machine learning based approach to a situation where a website has no specific data related to user preferences. Instead, the browsing patterns of previous users on a website are observed and analyzed to provide an estimation of the possible interests of new users.
The K-means clustering algorithm is used to evaluate the validity of clustering anonymous session data from a website visitor tracking system as a basis for identifying types of users. The clustered data is plotted into a scatterplot for analysis and used to examine the existence of clusters. In the case of distinct clustering, the clusters can be labeled as user groups, and further visitors can quickly be assigned to one of the groups to provide targeted content to.
K-means clustering is shown to perform suboptimally due to limitations in the algorithm’s implementation as well as high amounts of intracluster noise in the source data. However, the data exhibits areas of density and sparsity and could potentially provide meaningful results with a different clustering algorithm