Analyzing power efficiency of optimization techniques and algorithm design methods for applications on heterogeneous platforms

Abstract

Graphics processing units (GPUs) have become widely accepted as the computing platform of choice in many high performance computing domains. The availability of programming standards such as OpenCL are used to leverage the inherent parallelism offered by GPUs. Source code optimizations such as loop unrolling and tiling when targeted to heterogeneous applications have reported large gains in performance. However, given the power consumption of GPUs, platforms can exhaust their power budgets quickly. Better solutions are needed to effectively exploit the power-efficiency available on heterogeneous systems. In this work, we evaluate the power/performance efficiency of different optimizations used on heterogeneous applications. We analyze the power/performance trade-off by evaluating energy consumption of the optimizations. We compare the performance of different optimization techniques on four different fast Fourier transform implementations. Our study covers discrete GPUs, shared memory GPUs (APUs) and low power system-on-chip (SoC) devices, and includes hardware from AMD (Llano APUs and the Southern Islands GPU), Nvidia (Kepler), Intel (Ivy Bridge) and Qualcomm (Snapdragon S4) as test platforms. The study identifies the architectural and algorithmic factors which can most impact power consumption. We explore a range of application optimizations which show an increase in power consumption by 27%, but result in more than 1.8 × increase in speed of performance. We observe up to an 18% reduction in power consumption due to reduced kernel calls across FFT implementations. We also observe an 11% variation in energy consumption among different optimizations. We highlight how different optimizations can improve the execution performance of a heterogeneous application, but also impact the power efficiency of the application. More importantly, we demonstrate that different algorithms implementing the same fundamental function (FFT) can perform with vast differences based on the target hardware and associated application design.

Keywords

OpenCL fast Fourier Transform power optimizations GPUs system-on-chip

Get full access to this article

View all access options for this article.

References

AMD (n.d.) AMD SDK (formerly ATI Stream). Available at: http://developer.amd.com/gpu/AMDAPPSDK/.

AMD (n.d.) clAmdfft, OpenCL FFT library from AMD. Available at: http://www.bealto.com/gpu-fft.html.

Apple (n.d.) Apple implementation of FFT using OpenCL. Available at: https://developer.apple.com/library/mac/samplecode/OpenCL_FFT/Introduction/Intro.html

Bakhoda

Yuan

Fung

WWL

. (2009) Analyzing CUDA workloads using a detailed GPU simulator. In: Proceedings of 2009 IEEE international symposium on performance analysis of systems and software, Boston, MA, 26–28 April 2009, pp.163–174.

Boudier

Sellers

(2011) Memory system on Fusion APUs. In: AMD Fusion Developer Summit, Bellevue, Washington, USA, 13–16 June 2011.

BSquare (n.d.) Qualcomm Snapdragon S4 Pro APQ8064 MDP tablet datasheet. Available at: http://www.bsquare.com/Documents/APQ8064%20MDP%20Tablet.pdf.

Collange

Defour

Tisserand

(2009) Power consumption of GPUs from a software perspective. In: Computational Science—ICCS 2009 (Lecture Notes in Computer Science, Vol. 5544). Berlin; Heidelberg, Germany: Springer, pp.914–923.

Damaraju

George

Jahagirdar

. (2012) A 22 nm IA multi-CPU and GPU system-on-chip. In: Proceedings of 2012 IEEE international solid-state circuits conference, San Francisco, CA, 19–23 February 2012, pp.56–57.

Danalis

Marin

McCurdy

. (2010) The Scalable HeterOgeneous Computing (SHOC) Benchmark Suite. Oak Ridge, TN: Oak Ridge National Laboratory. Distributed by the Office of Scientific and Technical Information, U.S. Dept. of Energy.

10.

Deschizeaux

Blanc

(2007) Imaging earth's subsurface using CUDA. GPU Gems3: 831–850.

11.

Duhamel

Vetterli

(1990) Fast Fourier transforms: A tutorial review and a state of the art. Signal Processing19(4): 259–299. doi: 10.1016/0165-1684(90)90158-U.

12.

Frigo

Johnson

(2005) The Design and Implementation of FFTW3. Proceedings of the IEEE93(2): 216–231. doi: 10.1109/JPROC.2004.840301.

13.

Gaster

Howes

Kaeli

. (2011) Heterogeneous Computing with OpenCL. Waltham, MA: Morgan Kaufmann, 2011.

14.

Govindaraju

Lloyd

Dotsenko

. (2008) High performance discrete Fourier transforms on graphics processors. In: Proceedings of SC 2008—International conference for high performance computing, networking, storage and analysis, Austin, TX, 15–21 November 2008, pp.1–12.

15.

Hassanieh

Adib

Katabi

. (2012) Faster GPS via the sparse Fourier transform. In: Proceedings of the 18th annual international conference on mobile computing and networking—MOBICOM ’12, Istanbul, Turkey, p.353–364.

16.

Hong

Kim

(2009) An analytical model for a GPU architecture with memory-level and thread-level parallelism awareness. ACM SIGARCH Computer Architecture News37(3): 152. doi: 10.1145/1555815.1555775.

17.

Jang

Pien

. (2009) Architecture-aware optimization targeting multithreaded stream computing. In: Proceedings of 2nd workshop on general purpose processing on graphics processing units—GPGPU-2, Washington DC, USA, 8 March 2009, pp.62–70. doi: 10.1145/1513895.1513903.

18.

Jang

Schaa

Mistry

. (2011) Exploiting memory access patterns to improve memory performance in data-parallel architectures. IEEE Transactions on Parallel and Distributed Systems22(1): 105–118. doi: 10.1109/TPDS.2010.107.

19.

Kasichayanula

Terpstra

Luszczek

. (2012) Power aware computing on GPUs. In: Proceedings of symposium of application accelerators in high performance computing—SAAHPC ‘12, Lemont, Illinois, USA, July 2012, pp.63–74.

20.

Liu

Zhang

Shen

(2009) A cross-input adaptive framework for GPU program optimizations. In: Proceedings of IEEE international symposium on parallel and distributed processing (IPDPS 2009), Rome, Italy, May 2009, pp.1–10.

21.

Mantor

Houston

(2011) AMD graphic core next, low power high performance graphics and parallel compute. In: AMD Fusion Developer Summit, Bellevue, Washington, USA, 13–16 June 2011.

22.

Nagasaka

Maruyama

Nukada

. (2010) Statistical power modeling of GPU kernels using performance counters. In: Proceedings of IEEE international conference on green computing, Chicago, IL, 15–18 August 2010, pp.115–122.

23.

Nickolls

Dally

(2010) The GPU computing era. IEEE Micro30(2): 56–69. doi: 10.1109/MM.2010.41.

24.

Nukada

Ogata

Endo

. (2008) Bandwidth intensive 3-D FFT kernel for GPUs using CUDA. In: Proceedings of SC 2008—International conference for high performance computing, networking, storage and analysis, Austin, TX, 15–21 November 2008, pp.1–11.

25.

Nvidia (2010) Cufft library. Available at: https://developer.nvidia.com/cufft.

26.

Nvidia (2012) Whitepaper on NVIDIA GeForce GTX 680. Technical report.

27.

Nvidia (n.d.) Nvidia management library (NVML) Available at: http://developer.nvidia.com/cuda/nvidia-management-library-nvml.

28.

Owens

Houston

Luebke

. (2008) GPU computing. Proceedings of the IEEE96(5): 879–899.

29.

Pharr

Fernando

(2005) GPU Gems 2:Programming Techniques For High-Performance Graphics And General-Purpose Computation. Upper Saddle River, NJ: Addison-Wesley.

30.

Rivera

Schaa

Moffie

. (2007) Exploring novel parallelization technologies for 3-D imaging applications. In: Proceedings of 19th international symposium on computer architecture and high performance computing (SBAC-PAD’07), Gramado, Brazil, October 2007, pp.26–33.

31.

Rofouei

Stathopoulos

Ryffel

. (2008) Energy-aware high performance computing with graphic processing units. In: Proceedings of the 2008 conference on power aware computing and systems, San Diego, California, USA, December 2008. Available at: https://www.usenix.org/conference/hotpower-08/energy-aware-high-performance-computing-graphic-processing-units.

32.

Ryoo

Rodrigues

Baghsorkhi

. (2008) Optimization principles and application performance evaluation of a multithreaded GPU using CUDA. In: Proceedings of the 13th ACM SIGPLAN symposium on principles and practice of parallel programming (PPoPP ’08), Salt Lake City, Utah, USA, February 2008, pp.73–82.

33.

Spafford

Meredith

Lee

. (2012) The tradeoffs of fused memory hierarchies in heterogeneous computing architectures. In: Proceedings of the 9th conference on computing frontiers (CF ’12), New York, May 2012, p.103–112.

34.

Stone

Haldar

. (2008) How GPUs can improve the quality of magnetic resonance imaging. Urbana51: 61801.

35.

Suda

Ren

(2009) Accurate measurements and precise modeling of power dissipation of CUDA kernels toward power optimized high performance CPU-GPU computing. In: 2009 international conference on parallel and distributed computing, applications and technologies, Hiroshima, Japan, December 2009, pp.432–438.

36.

Tsoi

Luk

(2011) Power profiling and optimization for heterogeneous multi-core systems. ACM SIGARCH Computer Architecture News39(4): 8. doi: 10.1145/2082156.2082159.

37.

Ubal

Mistry

Schaa

. (2012) Multi2Sim: A simulation framework for CPU-GPU computing. In: Proceedings of the 21st international conference on parallel architectures and compilation techniques, Minneapolis, USA, September 2012, pp.335–344.

38.

Ukidave

Ziabari

Mistry

. (2013) Quantifying energy efficiency of FFT on heterogeneous platforms. In: Proceedings of international symposium on performance analysis of systems and software (ISPASS 2013), Austin, Texas, USA, April 2013, pp.235–244.

39.

Volkov

Kazian

(2008) Fitting FFT onto the G80 architecture. Available at: http://www.cs.berkeley.edu/kubitron/courses/cs258-S08/projects/reports/project6_report.pdf.

40.

Williams

Waterman

Patterson

(2009) Roofline: An insightful visual performance model for multicore architectures. Communications of the ACM52(4): 65–76.

41.

Vijaykrishnan

Kandemir

. (2000) The design and use of simplepower. In: Proceedings of the 37th conference on design automation (DAC ’00), New York, June 2000, pp.340–345.