prikaz prve stranice dokumenta Usporedba korištenja različitih vrsta računalnih resursa za simulaciju neuronskih mreža
Access restricted to authenticated users
master's thesis
Usporedba korištenja različitih vrsta računalnih resursa za simulaciju neuronskih mreža
2021. urn:nbn:hr:179:228992
Rančić, Luka
University of Split
Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture
Department of Electronics and Computing

Cite this document

Rančić, L. (2021). Usporedba korištenja različitih vrsta računalnih resursa za simulaciju neuronskih mreža (Master's thesis). Split: University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture. Retrieved from https://urn.nsk.hr/urn:nbn:hr:179:228992

Rančić, Luka. "Usporedba korištenja različitih vrsta računalnih resursa za simulaciju neuronskih mreža." Master's thesis, University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, 2021. https://urn.nsk.hr/urn:nbn:hr:179:228992

Rančić, Luka. "Usporedba korištenja različitih vrsta računalnih resursa za simulaciju neuronskih mreža." Master's thesis, University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, 2021. https://urn.nsk.hr/urn:nbn:hr:179:228992

Rančić, L. (2021). 'Usporedba korištenja različitih vrsta računalnih resursa za simulaciju neuronskih mreža', Master's thesis, University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, accessed 02 January 2025, https://urn.nsk.hr/urn:nbn:hr:179:228992

Rančić L. Usporedba korištenja različitih vrsta računalnih resursa za simulaciju neuronskih mreža [Master's thesis]. Split: University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture; 2021 [cited 2025 January 02] Available at: https://urn.nsk.hr/urn:nbn:hr:179:228992

L. Rančić, "Usporedba korištenja različitih vrsta računalnih resursa za simulaciju neuronskih mreža", Master's thesis, University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, Split, 2021. Available at: https://urn.nsk.hr/urn:nbn:hr:179:228992

Please login to the repository to save this object to your list.
Metadata
TitleUsporedba korištenja različitih vrsta računalnih resursa za simulaciju neuronskih mreža
Title (english)Comparison of the use of different types of computing resources for simulation of neural networks
AuthorLuka Rančić
MentorEugen Mudnić (mentor)
GranterUniversity of Split
Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture
(Department of Electronics and Computing)
Split
Defense date and country2021-03-03, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Abstract
U ovom radu prikazana je jednostavnost korištenja i prednosti Amazon web usluga te razlike u performansama izvršavanja testova iz područja dubokog učenja u različitim okolinama i na računalima različitih tehničkih karakteristika. Da bi se tema rada stavila u širi kontekst, opisani su osnovni koncepti i karakteristike računarstva u oblaku te pregled njegova povijesnog razvoja. Modeli računarstva u oblaku prema načinu implementacije i vrsti usluge predstavljaju temelje za razumijevanje važnosti korištenja usluga u oblaku. Također, opisani su osnovni principi i značajke Amazon web usluga, kao i njihov povijesni razvoj kako bi se lakše stekao realni dojam što one predstavljaju danas. Naglasak je stavljen na globalnu arhitekturu koja je osnova svih karakteristika koje obilježavaju Amazon web usluge, kao što su visoka dostupnost i performanse. Nadalje, prikazana je podjela Amazon web usluga na kategorije prema vrsti resursa koje pružaju korisnicima, a neke od kategorija usluga koje su opisane su: usluge poslužitelja, usluge za pohranu podataka, usluge za umrežavanje, usluge za rad s bazama podataka, usluge za sigurnost i usklađenost, analitičke usluge te usluge za upravljanje i administrativne poslove. Detaljno su opisane najznačajnije i najčešće korištene usluge svake od kategorija. Nakon toga, istaknute su karakteristike i specifikacije EC2 instanci i njihova podjela na tipove ovisno imaju li jedinstvenu namjenu ili su predviđene za opću uporabu. Pobliže su opisane P3 EC2 i G4dn EC2 instance koje su namijenjene za razvoj sustava visokih grafičkih zahtjeva, uz poseban osvrt na korištenje tih instanci u područjima strojnog i dubokog učenja.Također, definirane su prednosti korištenja AMI-ja za duboko učenje (eng. Deep Learning AMI). Jednostavnost pokretanja i korištenja EC2 instance potkrijepljena je detaljno opisanim postupkom podizanja AWS EC2 C5.large instance uz Deep Learning AMI. Na kraju, prikazani su rezultati izvršavanja testova obuke iz područja dubokog učenja u različitim okolinama i narazličitim računalima, a to su redom: računalo koje koristi GTX 1060 grafičku karticu, računalo koje koristi GTX 1080 grafičku karticu, AWS EC2 G4dn instanca te AWS EC2 C5.large instanca. Prikazane su razlike među pojedinim načinima izvršavanja te je donesen zaključak na temelju dobivenih rezultata.
Abstract (english)
This master's thesis presents the ease and benefits of using Amazon web services,as well as the differences in the performance of deep learning tests in different environments and on the computers of different technical characteristics. In order to put the topic of the master's thesis in a broader context, the basic concepts and characteristics of cloud computing are described, as well as an overview of its historical development. Cloud computing models according to the method of implementation and type of service are the basis for understanding the importance of using cloud services. Also, the fundamental principles and features of Amazon web services are described, as well as their historical development to make it easier to get a realistic impression of what they represent today. Emphasis is placed on the global architecture that underpins all the features that characterize Amazon web services, such as high availability and performance. Furthermore, the division of Amazon web services into categories according to the type of resources they provide is shown, and some of the service categories described are: compute services, storage services, networking services, database services, security and compliance services, analytics services and the management services. The most important and most frequently used services of each of the categories are described in detail. After that, the characteristics and specifications of EC2 instances and their division into types depending on whether they have single purpose or are intended for general use are highlighted. The EC2 P3 and EC2 G4dn instances intended for the development of high graphics requirements systems are described in detail, with special reference to the use of these instances in the areas of machine and deep learning. Also, the advantages of using AMI for Deep Learning (Deep Learning AMI) are defined. The ease of starting and using the EC2 instances is supported by the detailed procedure for launching the AWS EC2 C5.large instance with Deep Learning AMI. Finally, the results of performing Deep Learning training tests in different environments and on the different computers are presented, and those are: computer using GTX 1060 graphics card, computer using GTX 1080 graphics card, AWS EC2 G4dn instance and AWS EC2 C5.large instance. The differences between individual methods of execution are presented and a conclusion is made based on the obtained results.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:179:228992
Study programmeTitle: Computing
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsAccess restricted to authenticated users
Terms of useLicence In copyright icon
Created on2021-11-23 09:01:08