• Cite Icon1
  • https://doi.org/10.1145/3304108Copy DOI Icon

MiC

Show More
  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

Graphics processing units (GPUs) 1 have enjoyed increasing popularity in recent years, which benefits from, for example, general-purpose GPU (GPGPU) for parallel programs and new computing paradigms, such as the Internet of Things (IoT). GPUs hold great potential in providing effective solutions for big data analytics while the demands for processing large quantities of data in real time are also increasing. However, the pervasive presence of GPUs on mobile devices presents great challenges for GPGPU, mainly because GPGPU integrates a large amount of processor arrays and concurrent executing threads (up to hundreds of thousands). In particular, the root causes of performance loss in a GPGPU program can not be revealed in detail by current approaches. In this article, we propose MiC (Multi-level Characterization), a framework that comprehensively characterizes GPGPU kernels at the instruction, Basic Block (BBL), and thread levels. Specifically, we devise Instruction Vectors (IV) and Basic Blocks Vectors (BBV), a Thread Similarity Matrix (TSM), and a Divergence Flow Statistics Graph (DFSG) to profile information in each level. We use MiC to provide insights into GPGPU kernels through the characterizations of 34 kernels from popular GPGPU benchmark suites such as Compute Unified Device Architecture (CUDA) Software Development Kit (SDK), Rodinia, and Parboil. In comparison with Central Processing Unit (CPU) workloads, we conclude the key findings as follows: (1) There are comparable Instruction-Level Parallelism (ILP); (2) The BBL count is significantly smaller than CPU workloads—only 22.8 on average; (3) The dynamic instruction count per thread varies from dozens to tens of thousands and it is extremely small compared to CPU benchmarks; (4) The Pareto principle (also called 90/10 rule) does not apply to GPGPU kernels while it pervasively exists in CPU programs; (5) The loop patterns are dramatically different from those in CPU workloads; (6) The branch ratio is lower than that of CPU programs but higher than pure GPU workloads. In addition, we have also shown how TSM and DFSG are used to characterize the branch divergence in a visual way, to enable the analysis of thread behavior in GPGPU programs. In addition, we show an optimization case for a GPGPU kernel from the bottleneck identified through its characterization result, which improves 16.8% performance.

Similar Papers
  • Conference Article
  • Citations5

Application of GPGPU to Accelerate CFD Simulation

  • Jun 17, 2018
  • Shafiul A Mintu +1
  • Conference Article
  • Citations9

Database processing by Linear Regression on GPU using CUDA

  • Jul 01, 2011
  • Jyoti B Kulkarni +2
  • PDF
  • Research Article
  • Citations1

ORTHORECTIFICATION BY USING GPGPU METHOD

  • Jul 31, 2012
  • The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
  • H Sahin +1
  • Conference Article
  • Citations1

Improvement of Adaptive Fractal Image Coding Algorithm for GPGPU Systems Using Index Vectors

  • Mar 01, 2014
  • Akiyoshi Wakatani +1
  • Research Article

Analysis of Genetic Expression with Microarrays using GPU Implemented Algorithms

  • Sep 30, 2013
  • Computación y Sistemas
  • Eduardo Romero-Vivas +2
  • Research Article
  • Citations11

Energy Efficient Iris Recognition With Graphics Processing Units

  • Jan 01, 2016
  • IEEE Access
  • Ryan Rakvic +2
  • Book Chapter
  • Citations1

GPGPU-Based Painterly Rendering for Mobile Environment

  • Jan 01, 2017
  • Seulbeom Kim +2
  • Research Article

SU-E-T-28: Accelerated Event-By-Event Microdosimetry Monte Carlo Simulations of Low Energy Electron and Proton on a CUDA-Enabled GPU

  • Jun 01, 2012
  • Medical Physics
  • G Kalantzis +3
  • PDF
  • Research Article
  • Citations6

Parallelizing Multiple Flow Accumulation Algorithm using CUDA and OpenACC

  • Sep 03, 2019
  • ISPRS International Journal of Geo-Information
  • Natalija Stojanovic +1
  • Research Article

Improving MSAProbs Algorithm performance and Parallel Computing using GPU

  • Apr 18, 2023
  • International Journal of Computer Applications
  • Sally Zaki El-Hadary +2
  • Research Article
  • Citations11

Mitigation technique for performance degradation of virtual machine owing to GPU pass-through in fog computing

  • Jun 01, 2018
  • Journal of Communications and Networks
  • Jihun Kang +1
  • Conference Article

An efficient multiview video coding plan based on heterogeneous computing system

  • Apr 01, 2012
  • Jiangyu Sun +1
  • Research Article
  • Citations6

Implementation of Image Enhancement Algorithms and Recursive Ray Tracing using CUDA

  • Jan 01, 2016
  • Procedia Computer Science
  • Mr Diptarup Saha +3
  • PDF
  • Research Article
  • Citations3

General purpose graphic processing unit implementation of adaptive pulse compression algorithms

  • Aug 17, 2017
  • Journal of Applied Remote Sensing
  • Jingxiao Cai +1
  • Research Article
  • Citations30

Parallelization of the finite-difference time-domain method for room acoustics modelling based on CUDA

  • Dec 14, 2011
  • Mathematical and Computer Modelling
  • José J López +3
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.