Sattarov, Mirzabek and Kubayev, Saidazim (2026) TAQSIMLANGAN MUHITDA BLOKLI PARALLELLASHTIRILGAN INTELLEKTUAL TAHLIL ALGORITMLARINING SAMARADORLIGINI EKSPERIMENTAL BAHOLASH. Techscience.uz - Texnika fanlarining dolzarb masalalari, 4 (6). pp. 31-43. ISSN 3030-3702

[thumbnail of 4.+Sattarov+Mirzabek,+Kubayev+Saidazim.pdf] Text
4.+Sattarov+Mirzabek,+Kubayev+Saidazim.pdf - Published Version
Download (479kB)

Abstract

Ushbu maqolada taqsimlangan hisoblash muhitida blokli parallellashtirilgan intellektual tahlil algoritmlarining samaradorligini eksperimental baholash masalasi ko‘rib chiqiladi. Tadqiqotda K-means, DBSCAN va Random Forest algoritmlari tanlanib, ularning CPU’da ketma-ket bajarilishi, GPU’da parallel bajarilishi hamda blokli GPU-parallel bajarilish variantlari solishtirildi. Hisoblash eksperimenti Ubuntu operatsion tizimi, NVIDIA Tesla T4 15 GB grafik protsessori va 12 GB tezkor xotiraga ega muhitda amalga oshirildi. Klasterlash va klassifikatsiya masalalari uchun Covertype ma’lumotlar to‘plamidan foydalanildi. Natijalar GPU-parallel va blokli parallel yondashuvlar katta hajmli ma’lumotlarda bajarilish vaqtini sezilarli kamaytirishini ko‘rsatdi. Eng yuqori tezlashuv DBSCAN algoritmida 83martadan ortiq, K-means algoritmida 58 martadan ortiq va Random Forest algoritmida 21 martadan ortiq qiymatga yetdi.

Item Type: Article
Subjects: T Technology > T Technology (General)
Depositing User: Unnamed user with email info@ilmiykutubxona.uz
Date Deposited: 23 Jun 2026 00:52
Last Modified: 23 Jun 2026 00:52
URI: https://ilmiykutubxona.com/id/eprint/2152

Actions (login required)

View Item
View Item