Search

The finite-difference method for solving a nonlinear two-point boundary value problems consists of discretization and solving the resulting system of nonlinear equations, usually by the Newton’s method. In this paper, we modify Newton’s method to solve the system of nonlinear equations related to the discretization of a special class of the nonlinear two-point boundary value problems, where y′′is expressed as a multivariate polynomial in y and y′. In the proposed method, the new iteration point is determined as the minimal point of the square of 2-norm of residuals along Newton direction. The minimal point is computed using the solution of a univariate polynomial equation. The numerical examples show that the proposed method performs better than the standard Newton’s method with the unit step length.

Performance of memory allocation and garbage collection algorithms depends on many factors. These factors are mutually depedent and usually cannot be easily observed or measured since they are part of programmming language implementation. Therefore, we propose and implement a simulator, which enables the modelling and simulation of different memory-related processes. Organization of the simulator is described, followed by two examples of its usage for comparison of memory allocation and garbage collection algorithms.

A novel derivative-free algorithm for solving quasi-linear systems is presented. It resembles “classical” optimization approach but greatly simplifies computation, resulting in fast execution and numerical stability. Though the global convergence cannot be guaranteed, it turns out that the presented algorithm finds a solution as successfully as other commonly accepted methods. The algorithm is clearly developed and mathematically founded, and its properties are examined by comparisons with other methods.

One of the main principles of the smartphone and tablet operating systems is simplicity and accessibility. This feature was usually not available on the embedded systems and devices. By connecting these two types of devices, user interface and accessibility can be simplified. A myriad of new possibilities can be added to the embedded systems while giving smartphone and tablet users control over the external devices. The goal of this project is to create and build a fully functional model of a remotely controlled boat controllable through a mobile application. Connection is established through standard Bluetooth protocol and users are presented with the readings of the sensors found on the model. Boat is controlled by rotating the mobile device laterally and longitudinally and has the ability to change velocity or steer away from the obstacles.

Opis načina testiranja e-recpta, uspješnost testiranja i rad u stvarnoj okolini.

The aim of this paper is to propose various comparison criteria for reconfigurable processor architectures and to give an overview of the specification methods for them. The main contribution of this paper is a proposal of the new ontology-based representation of reconfigurable processor architectures.

In this paper two different methods for non-technical losses (NTL) detection are analyzed and new approach is proposed, based on the noticed drawbacks. It is shown that NTL can be successfully detected by a neural network trained by “artificial”, i.e., generated samples. This approach eliminates the need for many hard-to-obtain real life samples and the network can easily be trained to detect some new, nontypical occurrences in the system. This makes the proposed solution suitable for large companies that supply many different consumers who possibly change their consumption habits.

This paper describes mathematical modelling of traditional graphic structures within textile materials using binary matrices, which is made possible by the binary nature of these structures. An algorithm is presented for woven design construction (weave draft, threading draft, treadling draft and tie-up), as well as an algorithm for weave reinforcement using binary matrices. The general advantage of the algorithm is its simplicity and execution speed. Application of the weave reinforcement algorithm is also presented in this paper, and the computer program, written in C#, shows the ease of its implementation, opening the way to integration into existing and new CAD/CAM packages.

U radu je opisan vlastiti modul - aktivnost za podjelu zadataka koji je ugrađen u sustav za upravljanje učenjem Moodle. Aktivnost je prvenstveno namijenjena podjeli zadataka za laboratorijske vježbe, ali nije ograničena samo na tu namjenu već se može koristiti i za podjelu domaćih zadaća, zadataka za vježbu, seminara ili bilo kakvih drugih dokumenata namijenjenih polaznicima.

Prospective liberalization of European electrical energy market requires appropriate technological support because of increasing demands on energy quality control and measurement accuracy. This paper considers one possible direction of the development of highly accurate electronic energy meters based on spectral analysis. The concept is explained theoretically and justified by results obtained with a prototype. Important advantages and drawbacks are recognized, and directions for further development provided.

A simulation model of the radar network operation of five military radar stations has been developed. Simulation was performed in GPSS language and contains the time of operation of five radars through a period of one year, time of planned preventive maintenance, irregularities, time of corrective maintenance and maintenance teams. The simulation shows the influence of the number of maintenance teams on the availability of each radar and presents a good orienteering point for defining the optimal model of preventive and corrective maintenance of the radar network.

This paper describes the modelling language GPSS++. One of the main purposes of the language is its use in education. It extends the well-known discrete-event modelling language GPSS in several aspects, among which an object-oriented modelling is one of the most important. Other extensions of GPSS++ are: support for modular and hierarchical modelling, support for hybrid modelling (discrete and continuous), and significantly enhanced features for description of model behaviour (by using features of a general purpose programming languages). Syntax of the language is also modernized. In the same time, GPSS++ tries to retain the basic GPSS concept of modelling that uses transactions and blocks. The paper describes the main features of the GPSS++ mentioned above. Results of comparison of models written in GPSS and GPSS++ are given. The main advantages of GPSS++ over old versions of the language are better organization of the model and improved readability and writeability. Also, the language is better adapted to the programmers with the background in modern and widely used object-oriented languages. In addition, GPSS++ is better suited for larger models due to its modular and hierarchical organization. GPSS++ is currently under development and here we describe its first version and we also give directions for future improvements and research.

This paper describes the design and the implementation of a computer memory simulator which is modular, extensible and portable. It enables the simulation of various memory-related events from the aspect of the running program. The model includes the algorithms for memory allocation, garbage collection, stack behavior, etc. The simulation uses traces obtained by running the actual programs (benchmarks) written in any programming language. The simulator fulfills its main purpose: to simplify the development of algorithms for memory allocation and garbage collection. The algorithms can be easily modeled, validated, tested and compared in order to select the most efficient one.

This paper elaborates the present stage and predictable direction of future development of Sine Wave Calculator (SWC), a simple method for signal analysis in time domain that provides accuracy comparable to the best classical methods (relative errors about few times 10– 6). The main SWC advantages are extreme simplicity, possibility of selection of samples and straightforward calculations, which makes it preferable solution for quick implementation. The main shortage is still unknown mathematical analysis of errors, which are probably unequal for different signal parameters.

This paper elaborates the present stage and predictable direction of future development of Sine Wave Calculator (SWC), a simple method for signal analysis in time domain that provides accuracy comparable to the best classical methods (relative errors about few times 10– 6). The main SWC advantages are extreme simplicity and straightforward calculations, which makes it preferable solution for quick implementation. The main shortage is still unknown mathematical analysis of errors, which are probably unequal for different signal parameters.

The performance of a memory system depends to great extent on the algorithms used for memory allocation and garbage collection (GC). GC implementers need accurate data about memory objects' behaviour in order to produce high quality garbage collectors optimized for their particular problems. Therefore, different GC-related data should be collected and analysed afterwards. We propose a set of parameters that should be traced in benchmark programs and organization of an analysis environment. Then we present several examples of results obtainable by the proposed analysis environment. That results can be useful in improving the existing or in the creation of new GC algorithms. The profiled data are obtained by running benchmarks written in the Java programming language that we use as a representative of the languages that belong to the object oriented programming paradigm.

Extreme complexity of traffic systems imposes two problems in their simulation: the simulation speed, and the model's level of details. The simulation speed is improved by using better simulation algorithms, or by using more processing power. The level of details is managed by the most appropriate modeling scale. The simultaneous usage of multiple scales gives multiscale models. We have implemented a prototype simulator in order to investigate the possibility of using agents in multiscale modeling. The agents model characteristics of vehicles and drivers' behavior. Preliminary experiments have shown satisfactory simulation speed. We plan to enhance the simulator with agents that model road sections in order to support meso- and macroscale models.

Garbage collection (GC) algorithms are often based on beliefs about common behaviour of memory objects. Object behaviour patterns are very different because they depend on many factors (programming paradigm, language execution style, application itself, and programmer's style). There are several proposals and implementations of software for tracing the behaviour of memory objects or trace formats for heap allocation events (HATF).

The performance of memory system depends to great extent on the algorithms used for memory allocation and garbage collection (GC). GC implementers need accurate data about memory objects behaviour in order to produce high quality garbage collectors optimized for their particular problems. Therefore, different GC-related data should be collected and analysed afterwards. We propose a set of parameters that should be traced in benchmark programs and organization of an analysis environment. Then we present several examples of results that can be useful in improvements of performance of the memory system. Data are obtained by running Java benchmarks.

Ova disertacija prikazuje rezultate oblikovanja višeparadigmatičnog jezika namijenjenog modeliranju računalnog sklopovlja. Na početku su uvedni najosnovniji pojmovi iz područja modeliranja i simuliranja, s naglaskom na modeliranju i simuliranju računalnog sklopovlja. Dat je pregled programskih paradigmi i osnovnih značajki svake od njih. Opisana su svojstva jezika za modeliranje sklopovlja i istaknute njihove posebnosti u odnosu na opće programske jezike. Prikazani su najvažniji jezici za modeliranje sklopovlja koji su u praktičnoj upotrebi, kao i oni koji se koriste neimperativnim paradigmama. Procijenjene su moguće koristi od postojanja višeparadigmatičnosti u jezicima za opisivanje sklopovlja. Objašnjene su značajke koje bi programski jezici morali zadovoljavati i opća načela njihovog oblikovanja. Navedne su i obrazložene glavne smjernice i odluke donesene u vezi sa svojstvima i mogućnostima novog jezika koji se predlaže u ovoj disertaciji. Opisani su osnovni koncepti novog jezika nazvanog TL. On je oblikovan tako da uključuje svojstva imperativne, objektne i funkcijske paradigme, a također i dodatna svojstva potrebna za modeliranje računalnog sklopovlja. Sve najvažnije značajke jezika detaljno su objašnjene. Inače zasebni i neovisni pojmovi iz programskih jezika su poopćeni i ujedinjeni u pojam TL-objekta, kako bi se izbjeglo gomilanje jezičnih konstrukata iz različitih paradigmi. Brojni uobičajeni jezični konstrukti reinterpretirani su sa stajališta TL-objekta. Upotrebom TL-a izraženi su primjeri modela računalnog sklopovlja koji su uspoređeni s istovrijednim modelima u postoječim jezicima iste namjene. Kritički su razmotrena svojstva, prednosti i nedostatci novog jezika i predložene smjernice za daljnje istraživanje. This thesis presents the results of multiparadigm hardware description language design. At the beginning, the basics terms from the area of modelling and simulation are introduced, with emphasis on hardware modelling and simulation. The survey of the programming paradigms is given together with their main characteristics. Hardware description languages (HDL) and their features are described. The differences between HDLs and general purpose programming languages are stressed. The main HDLs in practical use are described. HDLs based on nonimperative paradigms are also described. Possible benefits of a multiparadigm approach in hardware modelling are estimated. The most important characteristics of programming languages, and the basic principles of programming language design are given. The main decisions concerning the features and possibilities of the new language are made. Ideas and concepts of the new language (named TL) are described. TL is multiparadigm language that incorporates imperative, object oriented, and functional paradigms. It also comprises all the necessary abilities for hardware description. The main language features of TL are presented in details. Notions that are usually separate in another programming languages are abstracted and unified into a single notion of TL-object. In that way the problem of accumulation of notions from different paradigms is avoided. Various conventional features are reinterpreted from the aspect of TL-objects. Several hardware descriptions are made by using TL. These models are compared with the adequate models written in the existing HDLs. The features, advantages, and disadvantages of TL are critically analysed and the possible directions for further improvements and research are given.

Ova disertacija prikazuje rezultate oblikovanja višeparadigmatičkog jezika namijenjenog modeliranju računalnog sklopovlja. Na početku su uvedeni najosnovniji pojmovi iz područja modeliranja i simuliranja, s naglaskom na modeliranju i simuliranju računalnog sklopovlja. Dat je pregled programskih paradigmi i osnovnih značajki svake od njih. Opisana su svojstva jezika za modeliranje sklopovlja i istaknute njihove posebnosti u odnosu na opće programske jezike. Prikazani su najvažniji jezici za modeliranje sklopovlja koji su u praktičnoj upotrebi, kao i oni koji se koriste neimperativnim paradigmama. Procijenjene su moguće koristi od postojanja više paradigmatičnosti u jezicima za opisivanja sklopovlja. Objašnjene su značajke koje bi programski jezici morali zadovoljavati i opća načela njihovog oblikovanja. Navedene su i obrazložene glavne smjernice i odluke donesene u vezi sa svojstvima i mogućnostima novog jezika koji se predlaže u ovoj disertaciji. Opisani su osnovni koncepti novog jezika nazvanog TL. On je oblikovan tako da uključuje svojstva imperativne, objektne i funkcijske paradigme, a također i dodatna svojstva potrebna za modeliranje računalnog sklopovlja. Sve najvažnije značajke jezika detaljno su objašnjene. Inače zasebni i neovisni pojmovi iz programskih jezika su poopćeni i ujedinjeni u pojam TL-objekta, kako bi se izbjeglo gomilanje jezičnih konstrukata iz različitih paradigmi. Brojni uobičajeni jezični konstrukti reinterpretirani su sa stajališta TL-objekta. Upotrebom TL-a izrađeni su primjeri modela računalnog sklopovlja koji su uspoređeni s istovrijednim modelima u postojećim jezicima iste namjene. Kritički su razmotrena svojstva, prednosti i nedostaci novog jezika i predložene smjernice za daljnje istraživanje.

The program package ATLAS (Advanced Tools and Languages for Microprocessor Architecture Simulation) consists of several tools. The most important are COMPAS (Configurable Microprocessor Architecture Simulator) and CONAS (Configurable Assembler).

Simulation methods are widely used in system analysis. An ATLAS digital simulation package is used to model and simulate complete DC motor drive control system. DC motor drive is modeled using transfer functions in Z domain. Controller of the drive is composed of a microprocessor with memory, D/A and A/D converter. Detailed model of a controller and satisfactory approximation of the drive makes possible testing and examining behavior of microprocessor controller. Model implemented in ATLAS enables the determination of optimal hardware and software solution for a desired control algorithm. Described model is extensively used for teaching principles of real time computer control.

This paper proposes and describes the method for performance evaluation and microprocessor bus allocation using a genetic algorithm

Predloženi su koncepti programskog okoliša za simulaciju arhitektura mikroračunala. Opisana je konkretna realizacija programskog okoliša i njegovih dijelova. Okoliš se sastoji od programskih alata i specifikacijskih jezika, koji omogućuju opis arhitekture mikroračunala na razini ponašanja s elementima strukturnog opisa. Razmatrani su i analizirani mogući postupci za izradu meta asemblera. Prikazana su svojstva realiziranog meta asemblera i meta jezika koji služi za opis asemblerskog jezika.

In this paper the digital system simulation program package ATLAS (Advanced Tools and Languages for microprocessor Architecture Simulation) is described. Different software modules, parts of ATLAS are developed: COMPAS - COnfigurable MicroProcessor Architecture Simulator, CONAS - CONfigurable ASsembler, ADEL - Assembler DEscription Language, COMDEL- COMponent DEscription Language and SYSDEL - SYStem DEscription Language. The simulator COMPAS and its implementation are explained in more details. The digital component description language features are given.

U svrhu simuliranja rada procesora, na Fakultetu elektrotehnike i računarstva razvijen je sustav zvan SSPARCSS. Sustav pruža mogućnost opisa modela arhitekture računala koristeći jezik COMDEL. Ovaj rad predlaže i implementira nekoliko promjena u SSPARCSS-u u svrhu olakšanja njegove uporabe. Uveden je novi koncept sabirnice koja grupira žice u zajedničku strukturu. Sabirnica omogućuje interakciju sa skupom žica preko grafičkog korisničkog sučelja, ali ne mijenja način izvođenja simulacije. Isto tako, u svrhu povezivanja izgleda i uporabe registra, razvijeni su novi prezentacijski elementi i naredbe COMDEL-a za prikaz registra. Izmjene skraćuju programski kod modela i pružaju jednostavan, standardizirani način upravljanja sabirnicom i prikazom registra. For the purpose of simulating processor operation, a system named SSPARCSS was developed at the Faculty of Electrical Engineering and Computing. The system provides the ability to define a model of a computer architecture using a language named COMDEL. This work proposes and implements several changes in SSPARCSS to make its usage easier. A new concept of a bus was introduced. The bus groupes wires into a single structure. The bus enables the interaction with wires through the graphical user interface, but does not change the way the simulation is run. Likewise, for the purpose of connecting the appearance and the usage of the register, new graphical components and COMDEL instructions for register display were developed. These changes shorten the program code for the model. The changes provide a simple, standardized method of management of buses and register depiction.

Diplomski rad opisuje izradu web-servisa koji optimizira dostavljanje transformiranog videosadržaja korisnicima te olakšavanja rada s videotransformacijama za dizajnere i programere prilikom izrade web-aplikacija. Prolazi se kroz problematiku, upoznavanje s videoformatima i codecima te pristup rješavanju problema kroz jezik Go uz pomoć ffmpeg programskog paketa. This masters' thesis tackles the problem of optimally delivering transformed video content to users and easing the process of transforming videos for web designers and programmers. Technical problems with such transformations are explained, video containers and codecs are introduced and one of possible solutions written in Go language with the help of ffmpeg software suite is presented.

Ovaj rad opisuje programski paket SSPARCSS i interni simulacijski jezik COMDEL. Definiran je i opisan COMDEL shema jezik (CSL) korišten u radu Shema editora. Implementirano je rješenje i opisana implementacija najvažnijih procesa editora. Posljednje poglavlje se bavi procesom validacije sheme i izradom COMDEL modela. This work describes software suite SSPARCSS and internally used simulation language COMDEL. COMDEL schema language (CSL) is defined, and description of the language is given. Solution is implemented and an implementation of the most important Schema editor processes are described. Last chapter defines the process of schema validation and COMDEL model generation.

U ovom radu opisana je ideja i implementacija vremenskog dijagrama stanja sabirnica, kao nadogradnja postojećeg simulatora modelacijskog jezika COMDEL2 u sklopu programskog sustava SSPARCSS. Prikaz stanja sabirnica iznimno je bitan za praćenje komunikacije između simuliranih računalnih komponenti, ali i za ispravljanje grešaka u modelu. U radu su redom navedeni i objašnjeni svi programski objekti pomoću kojih je izvedena detekcija i pohranjivanje promjena stanja sabirnica te prikaz tih promjena unutar grafičkog sučelja, u korelaciji s procesom za uzorkovanje. Poseban je naglasak stavljen na sinkronizaciju simulacije i prikaza te na algoritam animacije u vremenskom dijagramu. This paper describes the idea and implementation of the bus time diagram display as part of an upgrade of the existing simulator for COMDEL2 modeling language within SSPARCSS software system. Bus state display is extremely important for monitoring communication between the simulated computer components, but also for model debugging. This paper explains all programming objects necessary for performing detection and memorization of bus state changes, as well as for displaying those changes within the graphical interface, in correlation with the sampling process. Special emphasis is placed on synchronizing the simulation and display processes and on the time diagram animation algorithm.

U okviru rada opisujemo generalnu ideju simulacije arhitekture računala i fokusiramo se na sustav SSPARCSS. Promatramo povijest i sadašnji način rada sustava što prvenstveno uključuje njegov simulacijski algoritam. Ističemo razne probleme sadašnjeg pristupa te za njih predlažemo i u konačnici implementiramo rješenja. We describe the general idea of computer architecture simulation and focus on the SSPARCSS system. We detail the history and current workings of the system which mainly includes its simulation algorithm. After demonstrating various problems with the current approach we suggest and implement their solutions.

Cilj rada je pokazati na koji su se način postojeći kod i funkcionalnosti prepisale tako da koriste gotovo rješenje za efikasno iscrtavanje koje nudi Qt biblioteka za izradu grafičkih sučelja. Rad pokazuje što se moglo iskoristiti ponovno, a što se moglo izbaciti i zašto. The goal of this thesis is to show how has the current animation system in SSPARCSS Software Suite been replaced with another, more optimized one that uses Qt library's capabilities by going through the whole process of analyzing what can be reused and how can it be implemented effectively without changing much of the existing codebase.

Ovaj završni rad opisuje dodavanje prikaza sabirničkih vremenskih dijagrama u SSPARCSS simulator. Kratko je opisan programski paket SSPARCSS i njegov simulator. Opisana je osnovna građa i koncept dijela programa za prikaz sabirničkih dijagrama. Podsustav za prikaz je uspješno implementiran. Implementacija je objašnjena i uspoređena s početnim zahtjevima. Na kraju je dan zaključak i smjernice za mogući daljnji razvoj prikaza sabirničkih dijagrama. This graduation work describes the addition of bus timing diagrams in the SSPARCSS simulator. Software suite SSPARCSS and its simulator are briefly described. The application part for graphical presentation of bus diagrams is explained with the emphasis on its main concepts and structure. The presentation of bus diagrams is successfully implemented. The implementation is described and obtained results are compared with the given requirements. At the end, conclusion is given together with a few ideas for future work.

COMDEL je simulacijski jezik namijenjen za opis sučelja i ponašanja digitalnih komponenata. Za izvršavanje datoteka napisanih u COMDEL-u koristi se simulator COMPAS koji se sastoji od prevoditelja i izvršitelja. U procesu prevođenja najprije se pri parsiranju ulazne datoteke generiraju čvorovi apstraktnog sintaksnog stabla (AST). AST predstavlja polaznu strukturu nad kojom se zatim izvode postupci semantičke analize i generiranja bajtkoda. U jeziku COMDEL napravljeno je nekoliko izmjena: dodana je mogućnost korištenja naredbe switch uz predznačno i nepredznačno proširivanje brojeva, dodan je novi tip brojeva (x-brojevi) te je omogućena njihova upotreba u izrazima za usporedbu, uvedena je nova naredba decode koja je optimiziranija u odnosu na naredbu switch. Svaka navedena izmjena detaljno je opisana u kontekstu prethodno nabrojanih faza prevođenja jezika. COMDEL is a simulation language used for describing interfaces and behaviour of digital components. COMDEL files are processed by simulator COMPAS which consists both of compiler and executor. At the beginning of the compilation process, when parsing file, nodes of Abstract Syntax Tree (AST) are created. AST represents a key structure for semantic analysis and bytecode generation procedures. In COMDEL language, several modifications were made: added signed and unsigned form of switch statement, added a new type of numbers (x-numbers) adapted to use in comparison expression, introduced a new statement called decode, which executes more efficiently compared to switch statement. Each listed modification has been described in detail inside the context of the compilation phases mentioned earlier.

U ovom radu se sustavno opisuje dodavanje podrške za pronalaženje programskih grešaka u postojeći simulator arhitektura računala COMPAS. U prvom dijelu opisuju se promjene kroz sve slojeve programske podrške za simuliranje ciljne arhitekture. Promjene su napravljene u gramatici jezika COMDEL, izgradnji opisnog modela i prevođenju COMDEL programa u izvršni model. U drugom dijelu je prikazana izgradnja grafičkog korisničkog sučelja pomoću kojeg korisnik može koristiti podršku za pronalaženje programskih grešaka u mnemoničkim programima koji se simuliraju. This thesis describes adding of debugging support in the existing computer architecture simulator COMPAS. The first part of the thesis describes upgrades in core layers of COMPAS simulator. Upgrades have been made in COMDEL language grammar, building descriptive model support, and compiling COMDEL model into an executable model. The second part describes building graphical user interface through which user can use upgrades in core simulator part for debugging assembly files loaded into simulator.

Vizualni programski jezici često su zbog svojih paradigmi i značajki prilagođeni stručnjacima u neračunarskim područjima. Osmišljeni su tako da omoguće izražavanje vizualnim simbolima bliskima domeni za koju su namijenjeni, da budu jednostavni za upotrebu i da sakrivaju tehničke aspekte jezika od programera. U tom su smislu posebno zanimljivi i često korišteni vizualni jezici u području glazbe. Postojeći vizualni jezici u domeni glazbe koriste audio signale, odnosno matematičkim formulacijama opisan zvuk kao temeljni tip podataka. Međutim, rad s audio signalima zahtijeva poznavanje obrade signala i programerske vještine kako bi se na željeni način moglo njima upravljati. Glazbenicima ovo otežava korištenje jezika. U radu se stoga predlaže jezik koji se oslanja na koncepte timbralnih atributa kako bi se približio vokabularu i predznanju glazbenika. Jezik pritom pokušava zadržati osnovni koncept rada koji je glazbenicima poznat iz postojećih jezika. Kako bi se ostvario novi jezik, istraženi su različiti postupci preslikavanja timbralnih atributa u parametre audio sintetizatora. Zatim su timbralni atributi uklopljeni u tok podataka jezika sličan onome kod postojećih rješenja. Većina uobičajenih blokova i funkcionalnosti takvih jezika prilagođena je za rad s timbralnim atributima. Prilikom oblikovanja, validacije i evaluacije predloženog jezika korištene su metode kvantitativne i kvalitativne analize pomoću anketa te formalne analize pomoću kognitivnih dimenzija. Dodatno, predložena su specifična poboljšanja sučelja blokova u jeziku tako da olakšaju manipulaciju timbralnim atributima. Predstavljena su i rješenja za vizualizaciju rezultata programa i izlaznog zvuka u domeni timbralnih atributa. Konačno, predloženo je rješenje za vizualno uspoređivanja programa čiji je cilj lakše praćenje promjena među inačicama programa kao i otkrivanje pogrešaka. The paradigms and features of visual programming languages are often based on the needs of experts in fields outside computer science. They enable programming using visual symbols closely related to specific domains. Visual languages are prominent in the field of music. Existing visual languages for music use audio signals as their main data type. The manipulation of audio signals requires understanding of signal processing and programming which inhibits musicians from using these languages. This PhD thesis thus proposes a language that employs timbral attributes to bring programming closer to the vocabulary and knowledge of musicians. Simultaneously, the new language maintains existing best practices and basic principles. To model the new language, algorithms that translate timbral attributes into sound synthesis parameters were researched first. Then, timbral attributes were included in the language's data flow. Most of the usual objects and functionalities were adapted to work with timbral attributes. The process of modeling, validating, and evaluating the language was driven by quantitative and qualitative analysis based on questionnaires and formal analysis based on the cognitive dimensions framework. Improvements to object's visual interfaces are also introduced to simplify manipulation of timbral attributes. Similarly, methods of audio signal visualization in the domain of timbral attributes are proposed. Finally, a solution for visual comparisons of programs, that simplifies change tracking and error detection, is demonstrated. Scientific contributions: 1.Visual programming language for sound synthesis based on attribute flow and for computer music creation based on time-domain manipulations. 2.Models of exploration and visualization of sound synthesis results and for the intuitive control of a sound synthesizer and its parameters. 3. Improved and domain-adjusted methods for visual programs comparison.

Kao rezultat ovog rada nadograđen je virtualni stroj simulatora arhitekture procesora COMPAS 2, programskog sustava ATLAS 2. U virtualni stoj i izvršni model simulatora dodana je podrška za korištenje točaka prekida na način koji je opisan programskim jezikom COMDEL 2. U virtualnom stroju su popravljeni mehanizmi rezanja bitova i čišćenja okvira stoga pri povratku iz korisničkih funkcija. Implementirane su funkcije logički-i, logički-ili i funkcija logičke negacije, dodana je podrška za operacije množenja, dijeljenja i modulo brojeva s predznakom, te funkcija negacije bitova s postavljanjem pripadnih zastavica. U radu su opisani koncepti rada pojedinih dijelova sustava COMPAS i jezika COMPAS. The result of this thesis is an upgrade of the Virtual Machine in COMPAS 2 processor architecture simulator, a part of program package ATLAS 2. Support for using breakpoints, as described by the language COMDEL 2, was added to the Virtual Machine and the Executive model. Mechanisms for bit slicing and stack emptying in void function returns were fixed. Implementations were added for logic and, or and not functions, as well as signed multiplication, negation and modulo, and negation function with proper flag setting. Some basic concepts of the behavior of certain parts of COMPAS and its language, COMDEL, are described in this paper.

U diplomskom radu opisana je nadogradnja konfigurabilnog asemblerskog prevoditelja. U poglavlju 2 je detaljno opisan rad virtualnog stroja, pogotovo način na koji se izvode izrazi i naredbe te način na koji se pohranjuje kontekst tijekom izvođenja ADEL koda. Zatim su u poglavlju 3 opisane sve nadogradnje koje su implementirane. Navedene su glavne prednosti i koristi koje te nadogradnje donose, kao i način na koji su implementirane. Osim toga, u istom poglavlju opisane su i promjene koje su nastale kao rezultat ispravljanja grešaka iz prethodne verzije asemblerskog prevoditelja. Programsko rješenje implementirano je u jeziku C++. This graduation work describes software upgrade of configurable assembler. The work of virtual machine, especially the way in which expressions and ADEL commands are executed and run, is described in great details in section 2. The context of execution of ADEL code is also described in this section. All upgrades which are implemented in this graduation work are described in section 3. All the main advantages of those upgrades are described, along with the description of the implementation. Furthermore, all the changes made as a direct result of fixing the bugs from the last version of assembler are also described in section 3. The configurable assembler is implemented in C++ programming language.

Rezultat rada ovog projekta su nadogradnje simulatora COMPAS 2. Dodana je mogućnost zadavanja prekidnih točaka pomoću GUI-ja, dodana je mogućnost punjenja memorije sadržajem exe-datoteke odnosno strojnim kodom. Osim te dvije glavne nadogradnje implementirani su i razni popravci vezani za pogreške ili nelogičnosti u radu GUI-ja – osjetljivost polja s gumbima nakon greške u simulaciji, korištenje postojećeg display-prozora, a dodane su i neke nove stavke GUI-ju – statusna traka i čuvanje geometrije glavnog prozora. Rezultat rada je i stjecanje osnova za rad s Qt radnim okvirom kao i bolje poznavanje programskog jezika C++. Iako simulator već sada sadrži funkcionalnosti potrebne za rad s COMDEL-datotekama, potrebno je još par preinaka za lakše i ugodnije korištenje. Results of this paper are upgrades to COMPAS 2 simulator. Ability to add breakpoints through GUI is implemented as well as the ability to load contents of exe-file, that is, machine code in memory. Besides these two main upgrades, few issues and inconsistencies regarging GUI are fixed – sensitivity of buttons in toolbar after simulation error and usage of existing display-window, and some functions were added to GUI – status bar and keeping size of main window between application runs. “Side-result“ of this paper is also acquired knowledge of using Qt framework and better understanding of C++ programming language. Although simulator already has functions that are necessary to work with COMDEL files, few tweaks and upgrades are needed to make that work easier and more pleasant.

Memorija kao rijedak resurs u računalnim sustavima oduvijek je bila važna, a naročito u samim početcima računarstva kada je bila skupa i malih kapaciteta te se radi toga se javila potreba za racionalnim upravljanjem dostupnom memorijom. Kod zahtjevnijih i kompleksnijih računalnih programa programer jednostavno više ne može efikasno samostalno upravljati memorijom te je potrebna neka vrsta automatiziranog upravljanja memorijskim resursima. Upravo sakupljanje „smeća“ (eng. garbage collection) je dio programa koji se automatizirano brine o oslobađanju alociranih memorijskih resursa na gomili (eng. heap-u) pomoću određenih algoritama te određuje koji objekti nisu više dohvatljivi ili živi te samim time nepotrebni ili čak i smetnja normalnom odvijanju programa te ih oslobađa tj. oslobađa memorijski prostor koji su zauzimali. U ovom radu je pokazan rad Jikes RVM te njegovog modula za upravljanje memorijom MMTk pomoću kojeg možemo jednostavno i učinkovitu implementirati i testirati razne algoritme za učinkovito upravljanje memorijom. Također MMTk nam omogućuje testiranje algoritama koje koristi sam sakupljač smeća jer modul sadži već ugrađene tesntne skripte koje simuliraju razne situacije u kojima postoji potreba za uključivanje sakupljača smeća. U ovom radu sam testirao rad Mark Sweep algoritma kojeg koristi sakupljač smeća uz SegregatedFreeList algoritam za alokaciju. Memory as rare resource in computer sistems was always very important, especialy in the begging of computing when it was expensive and small capacities. Because of that we needed to racionlize our use of memory by better memory managment. With bigger and more comlex programs, programer can't mannage alone with all of memory an thats when we need some kind of automatized memory managment. Garbage collection is exactly that, a part of program that takes care of memory use on heap automaticly and with help of some alghotirams it can determine which objects are not reachable and not necessary or even disturbance to normaln program operation and it reallocates them and free the used memory space. In this work I've shown the basics of Jikes RVM and it's MMTk memory managment modul. With MMTk we can easily and efectivly implemment and test a various alghoritams for efficent memory mannagment. Also MMTk provides a test scripts which simulate differnt situations in which we could need a garbbage collector. In this work I tested work of Mark Sweep alghoritam which collector uses with Segregetad Free List alocation alghoritam.

Zadatak ovog rada je integracija modula za parsiranje ADEL datoteke i asemblera mnemoničke datoteke s grafičkim sučeljem, u cjelovitu aplikaciju konfigurabilnog asemblera. U drugom poglavlju je opisan Qt kao radno okruženje programskog jezika C++ i svojstva Qt-a koje su uveliko pojednostavile kreiranje grafičkog sučelja. U trećem poglavlju se nalazi kratak opis rada asemblera i novih funkcionalnosti dodanih u grafičko sučelje. U četvrtom poglavlju je opisana integracija modula kroz opisa razreda „MainProgram“ i njegovih metoda, zatim međusobne ovisnosti modula pri izvođenju ADEL parsera i asemblera mnemoničkih datoteka, te ispisivanje porukatijekom rada modula. The task of this bachelor thesis was integration of the ADEL parser and assembler modules with GUI in a complete configurable assembler application. In the second chapter Qt is described as a cross-platform framework for C++ programming language, together with Qt features that significantly simplified implementation of GUI. The third chapter describes the usage of configurable assembler and newly added functionalities in graphical user interface. The fourth chapter describes the integration of modules by describing class „MainProgram“ and its methods, interdependence between ADEL parser and assembler module, and printing out messages and errors that occur during the execution of the modules.

Cilj ovog završnog rada je bio napraviti alat za upravljanje dockerima s grafičkim korisničkim sučeljem. Danas su dockeri brzo rastuća tehnologija. Ipak, još uvijek je glavni način korištenja dockera isključivo preko ljuske operacijskog sustava. Jednostavno grafičko korisničko sučelje omogućuje prosječnom korisniku izbjegavanje učenja sintakse naredbi i interakcije naredbi te umjesto toga može odmah početi koristiti docker na koristan i produktivan način. Korisnicima koji već koriste docker tehnologiju ova aplikacija također ubrzava rad jer olakšava i pojednostavljuje kontrolu nad preslikama i kontejerima uklanjajući potrebu korisnika da mnoge stvari radi ručno, jednu po jednu. End result of this project was to create a docker management tool with graphical user interface. At the present moment, docker technology is advancing at an incredible rate. Nevertheless, the main way to access and use said technology is through the use of the shell. Simple graphical user interface enables the average user to avoid the need for learning all the commands and their syntax, and instead focus immediately on using docker productively. Experienced docker users will also find this tool helpful, for it removes the need of repetitive command executions using the pure text-based Linux shell.

Izrada standardne biblioteke za C++ sa naglaskom na efikasno upravljanje računalnim resursima i upotrebu naprednih svojstava modernih operativnih sustava. Razvoj modularne i skalabilne aplikacije za nenadgledanu reprodukciju audio i video sadržaja. Developing a standard library for C++ which aims to utilize computer resources most efficiently and take advantage of advanced features of the modern operating systems. Developing a modular and scalable application for unattended reproduction of audio and video materials.

U ovom diplomskom radu opisana je programska implementacija konfigurabilnog asemblerskog prevoditelja s podrškom za makro zamjene. Implementacija je ostvarena pomoću pet faza: učitavanje i predobrada, prvi makro prolazak, drugi makro prolazak, prvi prolazak asembliranja i drugi prolazak asembliranja. Izlazna datoteka generirana je samo ako svih pet faza prođe bez pogrešaka. U poglavlju 2 opisane su memorijske strukture podataka korištene u svrhu ostvarivanja rješenja. U poglavlju 3 opisane su sve faze i pojedine metode pomoću kojih se izvodi asembliranje. U poglavlju 4 opisan je postupak generiranja izlazne datoteke. U poglavlju 5 opisane su datoteke napisane u svrhu testiranja i organizacija istih. Konfigurabilni asemblerski prevoditelj implementiran je u programskom jeziku C++. Omogućen je rad na različitim ciljnim platformama. This graduate work describes software implementation of a configurable assembler with support for macros. Implementation is accomplished by five phases: loading and pretreatment, the first macro pass, the second macro pass, the first assembly pass and the second assembly pass. An output file is generated only if all five phases are completed without errors. The memory data structures used in order to achieve the solution are described in section 2. All the phases and various methods used for assembling are described in section 3. The process of generating an output file is described in section 4. The test files and their organization are described in section 5. The configurable assembler is implemented in C++ programming language. It can work on different target platforms.

U ovom završnom radu opisano je grafičko sučelje izrađeno za asemblerski prevoditelj procesora ARM i FRISC. Grafičko sučelje je programirano u programskom jeziku C++ korištenjem radnog okvira Qt za izradu grafičkog sučelja aplikacije. Sučelje sadrži integrirani tekstualni editor koji posjeduje brojač redaka te podržava sintaktičko bojenje prilagođeno vrsti procesora. Dodatno, tekstualni editor također podržava funkciju „Find & Replace“. Sintaktička bojenja opisana su u prilagođenim XML-datotekama. Sučelje omogućava otvaranje 10 prethodno otvorenih datoteka, te jednostavno učitavanje izvršne prevoditeljske datoteke kao i ADEL datoteke. Na početku je ukratko opisan radni okvir Qt u kojem je pisano grafičko sučelje. Zatim su opisane osnove rada asemblerskog prevoditelja za kojeg je pisano sučelje. Glavni dio rada sastoji se od opisa glavnih programskih komponenti sučelja- glavnih razreda i funkcija. This graduation work's goal was to develop Graphical User Interface which will support assembler application for procesors ARM and FRISC. GUI was made in C++, using Qt framework. This GUI contains integrated text editor with line numbering and syntax highlighting for each processor type. Text editor also supports Find & Replace functionality. Syntax highlighters are written in XML configuration files. GUI can also save up to 10 recently opened files, and makes it easy to load assembly and ADEL files necessary for compiling. This work contains short summary of Qt framework in which this GUI was written, basics of compiling process for programs written in assembly. Main part of this work contains descripitions of all the main classes and functions that are part of this application.

U završnom radu opisana je programska implementacija jezičnog prevoditelja za meta-jezik ADEL. Jezični prevoditelj prevodi ADEL datoteku u tri koraka: leksička, sintaksna i semantička analiza. U poglavlju 3 opisan je postupak leksičke analize za koji se koristi razred MyScanner i njegove funkcije. U poglavlju 4 opisan je postupak sintaksne analize i izgradnja apstraktnog sintaksnog stabla. U tom postupku se koristi razred MyParser i njegove funkcije, te ostali razredi izvedeni iz apstaktnog razreda AST koji stvaraju čvorove sintaksnog stabla. U poglavlju 5 opisan je postupak semantičke analize i povezivanje deklaracija s referencama na njih. U tom postupku razred MyChecker temeljan je razred. Jezični prevoditelj implementiran je u programskom jeziku C++. This graduation work describes implementation of compiler for ADEL metalanguage in software. Compiler compiles the ADEL file in three steps: lexical, syntax and semantics analysis. The process of lexical analysis is decribed in chapter 3. For that process class MyScanner is used, along with its functions. The process of syntax analysis and the process of building abstract syntax tree are described in chapter 4. Class MyParser, along with its functions, is used for this processes and classes derived from abstract AST class are used as elements of the abstract syntax tree. The process of semantics analysis and the process of connecting applied occurences with related declaration is described in chapter 5. Class MyChecker is the base class for this processes. Compiler is implemented in C++ programming language.

Analiza istraživačkog virtualnog stroja Jikes RVM i modula MMTk, izgradnja sakupljača označi-oslobodi u sklopu Jikes RVM-a i MMTk-a, te modifikacija tog sakupljača iz iterativnog u rekurzivni označi-oslobodi sakupljač. Analysis of the JikesRVM virtual machine and the Memory Management Toolkit MMTk, construction of a mark-sweep garbage collector in JikesRVM and MMTk, and the modification of the mark-sweep collector from an iterative to a recursive mark-sweep collector.

U ovom radu predstavljen je jezik PL kao i njegov prevoditelj. PL jezik je jednostavan programski jezik nalik poznatom jeziku C, ali bitno jednsotavniji. PL prevoditelj provodi fazu analize koja se sastoji od leksičke analize, sintaksne analize i semantičke analize. Svaka od analiza opisana je u poglavljima 2, 3 i 4 respektivno. Na kraju, dan je načelni izgled virtualnog stroja u poglavlju 5, kao i izgled međukoda koji PL prevoditelj generira za taj virtualni stroj. In this paper, we are building a compiler for a programming language PL (Project language) which is similar to a world-known programming language C, but far less complex. Compiler contains of three different types of analysis: lexical analysis (known as scanner), syntax analysis (known as parser) and contextual analysis, which are descibed in chapters 2, 3 and 4 respectively. In 5th chapter, we are describing a virtual machine which is supposed to run the bytecode that compiler provides. Also, bytecode itself is described within the same chapter.

Potrebno je definirati sustav za prikaz i obradu podataka te potom dizajnirati i izvesti predloženo rješenje za rad u stvarnom vremenu. Ostvariti komunikaciju između sustava za prikupljanje podataka i modula za prikaz podataka. Prilagoditi sustav za rad s podacima u mobilnom okruženju i udaljeno dijeljenje podataka. Omogućiti podršku za rad na više različitih razvojnih platformi. The aim of this thesis is to define a module for data processing and visualization. Module is to be designed and implemented to support real time operation. Communication between data collection and visualization units is to be established. System is configured in such a way that it operates in mobile environment and supports remote data distribution. System supports multiple development platforms and architectures.

Algoritmi za upravljanje memorijom uvelike utječu na brzinu rada programa i njegove memorijske zahtjeve pa su zbog toga predmet istraživanja već dugi niz godina. Iako postoji mnoštvo algoritama, oblikovanjem novih želi se postići efikasnije korištenje računalne memorije i poboljšati performanse rada sustava. Tradicionalno se algoritam za zbrinjavanje ugrađuje povrh algoritma za dodjelu. U ovom radu opisana je metoda kombiniranja mehanizama za dodjelu i mehanizama za zbrinjavanje memorije s ciljem oblikovanja jednoslojnog algoritma za upravljanje memorijom koji utjelovljuje najbolje karakteristike početnih algoritama, a istodobno primjenjuje nove paradigme. Model je zasnovan na iskustvu postojećih algoritama, a posebno na algoritmu za dodjelu s odijeljenim listama memorijskih blokova te klasičnom algoritmu označi-oslobodi za zbrinjavanje nedohvatljivih objekata. Jednoslojni algoritam ima manju složenost postupka zbrinjavanja u odnosu na algoritam označi-oslobodi i postiže veću efikasnost. Oblikovani algoritam evaluiran je simulacijama obzirom na performanse. On ima slične memorijske zahtjeve kao i klasični algoritam označi-oslobodi, no provodi 19% manje vremena u zbrinjavanju, a uz njega je sustav 28% brži. Temeljem analize rezultata rada ispitnih programa razvijeno je nekoliko modela i na osnovu njih oblikovano nekoliko inačica heurističkog mehanizma prilagodbe jednoslojnog algoritma značajkama memorijskih objekata, no u konačnoj evaluaciji se koristi ona inačica koji ostvaruje najprihvatljiviji omjer dobitka na brzini u odnosu na dodatno memorijsko zauzeće. Memorijski zahtjevi heurističkog jednoslojnog algoritma slični su zahtjevima standardnog jednoslojnog i klasičnog algoritma označi-oslobodi, međutim heuristički jednoslojni algoritam provodi 47% manje vremena u zbrinjavanju u odnosu na algoritam označi-oslobodi, dok je uz njega sustav u cjelini 30% brži. Rezultati ovog istraživanja pokazuju da se relativno jednostavnim, ali dobro osmišljenim mehanizmima i tehnikama mogu značajno poboljšati performanse memorijskih algoritama i time ukazuju na nekoliko mogućih smjerova daljnjih istraživanja. Memory management algorithms greatly affect the execution speed and memory requirements of programs which are executed. Although there are a number of existing algorithms, new ones are devised with the goal of achieving more efficient memory use and improved system performance. Traditionally a garbage collector is built in as a layer above a memory allocator. This research describes a method which combines memory allocation and garbage collection mechanisms to achieve a single layer method of memory management which embodies best characteristics of initial algorithms whilst simultaneously applying new paradigms. Resulting model is based on the well-established algorithms, in particular on memory allocation algorithm which uses segregated lists of free memory blocks and classical mark-and-sweep garbage collector. Analysis shows that the single layer algorithm has lower collection complexity and higher efficiency in comparison to the mark-and-sweep. Performances of a devised model are evaluated using simulation. Its memory requirements are similar to those of the classical mark-and-sweep but its garbage collection time and total time are improved by 19% and 28% respectively. Based on the analysis of the results, several models of heuristic mechanism which adapts to the behavior of memory objects are developed. However, in the final evaluation only the last model is considered as it shows the best memory-vs-speed trade-off. Memory consumption of the heuristic single layer algorithm is similar to those of the single layer algorithm and the classical mark-and-sweep. However, heuristic algorithm improves garbage collection time by 47% and total time by 30% in comparison to classical mark-and-sweep. This research shows that relatively simple, but well-devised mechanisms and techniques may greatly improve performances of memory algorithms thus suggesting several possible directions for further research.