Improving GPGPU concurrency with elastic kernels

Author:

Pai Sreepathi1,Thazhuthaveetil Matthew J.1,Govindarajan R.1

Affiliation:

1. Indian Institute of Science, Bangalore, India

Abstract

Each new generation of GPUs vastly increases the resources available to GPGPU programs. GPU programming models (like CUDA) were designed to scale to use these resources. However, we find that CUDA programs actually do not scale to utilize all available resources, with over 30% of resources going unused on average for programs of the Parboil2 suite that we used in our work. Current GPUs therefore allow concurrent execution of kernels to improve utilization. In this work, we study concurrent execution of GPU kernels using multiprogram workloads on current NVIDIA Fermi GPUs. On two-program workloads from the Parboil2 benchmark suite we find concurrent execution is often no better than serialized execution. We identify that the lack of control over resource allocation to kernels is a major serialization bottleneck. We propose transformations that convert CUDA kernels into elastic kernels which permit fine-grained control over their resource usage. We then propose several elastic-kernel aware concurrency policies that offer significantly better performance and concurrency compared to the current CUDA policy. We evaluate our proposals on real hardware using multiprogrammed workloads constructed from benchmarks in the Parboil 2 suite. On average, our proposals increase system throughput (STP) by 1.21x and improve the average normalized turnaround time (ANTT) by 3.73x for two-program workloads when compared to the current CUDA concurrency implementation.

Publisher

Association for Computing Machinery (ACM)

Subject

Computer Graphics and Computer-Aided Design,Software

Reference22 articles.

1. The case for GPGPU spatial multitasking

2. Analyzing CUDA workloads using a detailed GPU simulator

3. Rodinia: A benchmark suite for heterogeneous computing

4. M. Desnoyers etal LTTng-UST User Space Tracer. M. Desnoyers et al. LTTng-UST User Space Tracer.

Cited by 109 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. KeSCo: Compiler-based Kernel Scheduling for Multi-task GPU Applications;2023 IEEE 41st International Conference on Computer Design (ICCD);2023-11-06

2. Hierarchical Resource Partitioning on Modern GPUs: A Reinforcement Learning Approach;2023 IEEE International Conference on Cluster Computing (CLUSTER);2023-10-31

3. Optimum: Runtime optimization for multiple mixed model deployment deep learning inference;Journal of Systems Architecture;2023-08

4. DOPpler: Parallel Measurement Infrastructure for Auto-Tuning Deep Learning Tensor Programs;IEEE Transactions on Parallel and Distributed Systems;2023-07

5. ISPA: Exploiting Intra-SM Parallelism in GPUs via Fine-Grained Resource Management;IEEE Transactions on Computers;2023-05-01

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3