Browsing by Author "Oz, Isil"

Now showing 1 - 2 of 2

Community Detection for Large Graphs on GPUs With Unified Memory
(Institute of Electrical and Electronics Engineers Inc., 2024) Dincer, Emre; Öz, Işıl; Oz, Isil; 03.04. Department of Computer Engineering; 03. Faculty of Engineering; 01. Izmir Institute of Technology
While GPUs accelerate applications from different domains with different characteristics, processing large datasets gets infeasible on target systems with limited device memory. Unified memory support makes it possible to work with data larger than available GPU memory. However, page migration overhead for executions with irregular memory access patterns, like graph processing workloads, induces severe performance degradation. While memory hints help to deal with page movements by keeping data in suitable memory spaces, coarse-grain configurations can still not avoid migrations for executions having diverse data structures. In this work, we target the state-of-the-art CUDA implementation of the Louvain community detection algorithm and evaluate the impacts of the fine-grained unified memory hints on the performance. Our experimental evaluation shows that memory hints configured for specific data structures reveal significant performance improvements and enable us to work efficiently with large graphs.
Teaching Accelerated Computing With Hands-On Experience
(Institute of Electrical and Electronics Engineers Inc., 2025) Oz, Isil; 01. Izmir Institute of Technology; 03. Faculty of Engineering; 03.04. Department of Computer Engineering
Heterogeneous computing systems maintain high-performance executions with parallel hardware resources. Graphics Processing Units (GPUs) with many parallel efficient cores and high-bandwidth memory structures enable accelerated computing for high-performance, deep learning, and embedded programs from diverse domains. The expertise in GPU programming requires a significant effort to utilize parallel computational units efficiently. Teaching programming for heterogeneous systems also becomes difficult due to dedicated hardware requirements and up-to-date course materials. In this paper, we present our teaching experience in an undergraduate parallel programming course, where we adopt NVIDIA Deep Learning Institute workshop and teaching kit contents and GPU devices at different scales to expose students to a set of hardware platforms with hands-on coding experience. © 2025 Elsevier B.V., All rights reserved.