Characterizing Optimizations to Memory Access Patterns using Architecture-Independent Program Features
Article No.: 15, Pages 1 - 11
Abstract
High-performance computing developers are faced with the challenge of optimizing the performance of OpenCL workloads on diverse architectures. The Architecture-Independent Workload Characterization (AIWC) tool is a plugin for the Oclgrind OpenCL simulator that gathers metrics of OpenCL programs that can be used to understand and predict program performance on an arbitrary given hardware architecture. However, AIWC metrics are not always easily interpreted and do not reflect some important memory access patterns affecting efficiency across architectures. We propose a new metric of parallel spatial locality - the closeness of memory accesses simultaneously issued by OpenCL work-items (threads). We implement the parallel spatial locality metric in the AIWC framework and analyse gathered results on matrix multiply and the Extended OpenDwarfs OpenCL benchmarks. The differences in the observed parallel spatial locality metric across implementations of matrix multiply reflect the optimizations performed. The new metric can be used to distinguish between the OpenDwarfs benchmarks based on the memory access patterns affecting their performance on various architectures. The improvements suggested to AIWC will help HPC developers better understand memory access patterns of complex codes and guide optimization of codes for arbitrary hardware targets.
References
[1]
Extended OpenDwarfs. https://github.com/ANU-HPC/OpenDwarfs/commit/dee488cac9833f029dfada35ff6ae4077b68c4b5, Jan 2020.
[2]
Krste Asanovic, Ras Bodik, Bryan Christopher Catanzaro, Joseph James Gebis, Parry Husbands, Kurt Keutzer, David A Patterson, William Lester Plishker, John Shalf, Samuel Webb Williams, et al. The landscape of parallel computing research: A view from berkeley. Technical report, Technical Report UCB/EECS-2006-183, EECS Department, University of California, Berkeley, 2006.
[3]
Karthik Ganesan, Lizy John, Valentina Salapura, and James Sexton. A performance counter based workload characterization on Blue Gene/P. In International Conference on Parallel Processing (ICPP), pages 330--337. IEEE, 2008.
[4]
John L. Hennessy and David A. Patterson. Computer Architecture - A Quantitative Approach, 5th Edition. Morgan Kaufmann, 2012.
[5]
Kenneth Hoste and Lieven Eeckhout. Microarchitecture-independent Workload Characterization. IEEE Micro, 27(3), 2007.
[6]
Intel Corporation. Intel® 64 and IA-32 Architectures Optimization Reference Manual. 2016.
[7]
Beau Johnston and Josh Milthorpe. AIWC: OpenCL-based architecture-independent workload characterization. IEEE/ACM Workshop on the LLVM Compiler Infrastructure in HPC (LLVM-HPC), Nov 2018.
[8]
Beau Johnston and Josh Milthorpe. Dwarfs on accelerators: Enhancing OpenCL benchmarking for heterogeneous computing architectures. In Proceedings of the 47th International Conference on Parallel Processing Companion, ICPP '18, pages 4:1--4:10, New York, NY, USA, 2018. ACM.
[9]
Beau Johnston, James Price, Moritz Pflanzer, Petros Kalos, Tom Deakin, Nido Media, and Daniel Saier. BeauJoh/Oclgrind: Adding AIWC - An Architecture Independent Workload Characterisation Plugin. https://doi.org/10.5281/zenodo.1134175, December 2017.
[10]
Yooseong Kim and Aviral Shrivastava. CuMAPz: a tool to analyze memory access patterns in CUDA. In Leon Stok, Nikil D. Dutt, and Soha Hassoun, editors, Proceedings of the 48th Design Automation Conference, DAC 2011, San Diego, California, USA, June 5-10, 2011, pages 128--133. ACM, 2011.
[11]
Konstantinos Krommydas, Wu-chun Feng, Christos D Antonopoulos, and Nikolaos Bellas. OpenDwarfs: Characterization of dwarf-based benchmarks on fixed and reconfigurable architectures. Journal of Signal Processing Systems, 85(3):373--392, 2016.
[12]
Chi-Keung Luk, Robert Cohn, Robert Muth, Harish Patil, Artur Klauser, Geoff Lowney, Steven Wallace, Vijay Janapa Reddi, and Kim Hazelwood. Pin: building customized program analysis tools with dynamic instrumentation. In ACM SIGPLAN notices, volume 40, pages 190--200. ACM, 2005.
[13]
NVIDIA Corporation. CUDA C++ Best Practices Guide. 2019.
[14]
NVIDIA Corporation. CUDA C++ Programming Guide. 2019.
[15]
James Price and Simon McIntosh-Smith. Oclgrind: An extensible OpenCL device simulator. In Proceedings of the 3rd International Workshop on OpenCL, page 12. ACM, 2015.
[16]
Yakun Sophia Shao and David Brooks. ISA-independent workload characterization and its implications for specialized architectures. In IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS), pages 245--255. IEEE, 2013.
[17]
Kyle Spafford, Jeremy Meredith, and Jeffrey Vetter. Maestro: data orchestration and tuning for OpenCL devices. Euro-Par 2010-Parallel Processing, pages 275--286, 2010.
- Characterizing Optimizations to Memory Access Patterns using Architecture-Independent Program Features
Recommendations
Memory access patterns: the missing piece of the multi-GPU puzzle
SC '15: Proceedings of the International Conference for High Performance Computing, Networking, Storage and AnalysisWith the increased popularity of multi-GPU nodes in modern HPC clusters, it is imperative to develop matching programming paradigms for their efficient utilization. In order to take advantage of the local GPUs and the low-latency high-throughput ...
Enhancing Heterogeneous Computing Through OpenMP and GPU Graph
ICPP '24: Proceedings of the 53rd International Conference on Parallel ProcessingModern computing platforms are increasingly heterogeneous, most of them include accelerators such as GPU. OpenMP as the de-facto standard to parallelize CPU applications, incorporates target construct allowing users to offload work onto such ...
GPGPU-MiniBench: Accelerating GPGPU Micro-Architecture Simulation
Graphics processing units (GPU), due to their massive computational power with up to thousands of concurrent threads and general-purpose GPU (GPGPU) programming models such as CUDA and OpenCL, have opened up new opportunities for speeding up general-...
Comments
Please enable JavaScript to view thecomments powered by Disqus.Information & Contributors
Information
Published In
April 2020
104 pages
ISBN:9781450375313
DOI:10.1145/3388333
Copyright © 2020 ACM.
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from [email protected].
In-Cooperation
- Khronos: Khronos Group
- Codeplay: Codeplay Software Ltd.
- Intel: Intel
- The University of Bristol: The University of Bristol
- Tech Univ of Munich: Technical University of Munich
Publisher
Association for Computing Machinery
New York, NY, United States
Publication History
Published: 27 April 2020
Check for updates
Author Tags
Qualifiers
- Research-article
- Research
- Refereed limited
Conference
IWOCL '20
Acceptance Rates
IWOCL '20 Paper Acceptance Rate 21 of 30 submissions, 70%;
Overall Acceptance Rate 84 of 152 submissions, 55%
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 120Total Downloads
- Downloads (Last 12 months)4
- Downloads (Last 6 weeks)2
Reflects downloads up to 14 Feb 2025
Other Metrics
Citations
View Options
Login options
Check if you have access through your login credentials or your institution to get full access on this article.
Sign in