Raspodijeljena obrada velikih geoprostornih podataka na grafičkim procesorima

master's thesis

• Title (english): Distributed Processing of Big Geosptial Data on Graphics Processors
• Author: Nikola Radelić
• Mentor: Krešimir Pripužić (mentor)
• Committee member: Krešimir Pripužić (predsjednik povjerenstva)
• Committee member: Dragan Jevtić (član povjerenstva)
• Committee member: Martina Antonić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-07-11, Croatia

Abstract

Apache Spark je programska platforma za raspodijeljenu obradu velikih podataka na više računala. Trenutno je jedna od vodećih platformi otvorenog koda. Nvidia je 2020. godine predstavila akcelerator RAPIDS za platformu Spark koja omogućuje obradu podataka na grafičkim procesorima. Tvrtka Nvidia tvrdi ubrzanja za određene programe do 4 puta. U ovom radu istražena je platforma Apache Spark, akcelerator RAPIDS i knjižnica cuSpatial kako bi se provjerilo ubrzanje obrade podataka u usporedbi s tradicionalnim procesorima. Iz C++ knjižnice cuSpatial preko knjižnice JNI pozvana je nativna metoda koju Spark može koristiti kao SQL naredbu za obradu podataka. Pokazana su ubrzanja do 3.1 puta u usporedbi s obradom na procesoru.

Abstract (english)

Apache Spark is a software platform for distributed processing of big data across multiple computers. It is currently one of the leading open-source platforms. In 2020, Nvidia introduced the RAPIDS accelerator for the Spark platform, which enables data processing on graphics processors. Nvidia claims up to 4x speed improvements for certain programs. This paper explores the Apache Spark platform, the RAPIDS accelerator, and the cuSpatial library to verify data processing speedups compared to traditional processors. A native method from the C++ cuSpatial library was called via the JNI library, which Spark can use as an SQL command for data processing. Speedups of up to 3.1x were demonstrated compared to processing on a CPU.

• Keywords: spark; rapids; cuspatial; cuda; nvidia; apache; sql; udf; geoprostorni podatci; raspodijeljena obrada; jni
• Keywords (english): spark; rapids; cuspatial; cuda; nvidia; apache; sql; udf; geospatial data; distributed computing; jni
• Language: croatian
• URN:NBN: urn:nbn:hr:168:225560
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2024-10-12 21:58:00