Title: FPGA-based HPC accelerators: An evaluation on performance and energy efficiency

Journal Article · · Concurrency and Computation. Practice and Experience
DOI: https://doi.org/10.1002/cpe.6570 · OSTI ID:1959989
ORCiD logo [1];  [1];  [2]; ORCiD logo [1];  [1];  [1]
  1. Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United States)
  2. DEIB, Politecnico di Milano (Italy)
+ Show Author Affiliations

Hardware specialization is a promising direction for the future of digital computing. Reconfigurable technologies enable hardware specialization with modest non-recurring engineering cost, but their performance and energy efficiency compared to state-of-the-art processor architectures remain an open question. In this article, we use FPGAs to evaluate the benefits of building specialized hardware for numerical kernels found in scientific applications. In order to properly evaluate performance, we not only compare Intel Arria 10 and Xilinx U280 performance against Intel Xeon, Intel Xeon Phi, and NVIDIA V100 GPUs, but we also extend the Empirical Roofline Toolkit (ERT) to FPGAs in order to assess our results in terms of the Roofline model. We show design optimization and tuning techniques for peak FPGA performance at reasonable hardware usage and power consumption. As FPGA peak performance is known to be far less than that of a GPU, we also benchmark the energy efficiency of each platform for the scientific kernels comparing against microbenchmark and technological limits. Results show that while FPGAs struggle to compete in absolute terms with GPUs on memory- and compute-intensive kernels, they require far less power and can deliver nearly the same energy efficiency.

Research Organization:
Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA (United States)
Sponsoring Organization:
USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR)
Grant/Contract Number:
AC02-05CH11231
OSTI ID:
1959989
Journal Information:
Concurrency and Computation. Practice and Experience, Journal Name: Concurrency and Computation. Practice and Experience Journal Issue: 20 Vol. 34; ISSN 1532-0626
Publisher:
WileyCopyright Statement
Country of Publication:
United States
Language:
English

References (26)

Striped Smith-Waterman speeds database searches six times over other SIMD implementations journal November 2006
On the characterization of OpenCL dwarfs on fixed and reconfigurable platforms conference June 2014
Accelerating a Random Forest Classifier: Multi-Core, GP-GPU, or FPGA?
  • Van Essen, Brian; Macaraeg, Chris; Gokhale, Maya
  • 2012 IEEE 20th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM), 2012 IEEE 20th International Symposium on Field-Programmable Custom Computing Machines https://doi.org/10.1109/FCCM.2012.47
conference April 2012
Understanding Performance Differences of FPGAs and GPUs conference April 2018
Shuhai: Benchmarking High Bandwidth Memory On FPGAS conference May 2020
Performance comparison of FPGA, GPU and CPU in image processing conference August 2009
Trends of CPU, GPU and FPGA for high-performance computing conference September 2014
FPGA vs. GPU for sparse matrix vector multiply conference December 2009
Evaluating FPGA Accelerator Performance with a Parameterized OpenCL Adaptation of Selected Benchmarks of the HPCChallenge Benchmark Suite conference November 2020
Rodinia: A benchmark suite for heterogeneous computing conference October 2009
A Roofline Model of Energy
  • Choi, Jee Whan; Bedard, Daniel; Fowler, Robert
  • 2013 IEEE International Symposium on Parallel & Distributed Processing (IPDPS), 2013 IEEE 27th International Symposium on Parallel and Distributed Processing https://doi.org/10.1109/IPDPS.2013.77
conference May 2013
merAligner: A Fully Parallel Sequence Aligner conference May 2015
BLAS Comparison on FPGA, CPU and GPU conference July 2010
An Empirical Roofline Methodology for Quantitatively Assessing Performance Portability conference November 2018
The Performance and Energy Efficiency Potential of FPGAs in Scientific Computing conference November 2020
Accelerating Compute-Intensive Applications with GPUs and FPGAs conference June 2008
Evaluating and Optimizing OpenCL Kernels for High Performance Computing with FPGAs
  • Zohouri, Hamid Reza; Maruyama, Naoya; Smith, Aaron
  • SC16: International Conference for High Performance Computing, Networking, Storage and Analysis https://doi.org/10.1109/SC.2016.34
conference November 2016
Beyond the Roofline: Cache-Aware Power and Energy-Efficiency Modeling for Multi-Cores journal January 2017
Impact of modern memory subsystems on cache optimizations for stencil computations conference January 2005
Roofline: an insightful visual performance model for multicore architectures journal April 2009
Let there be light! conference June 2011
Can FPGAs Beat GPUs in Accelerating Next-Generation Deep Neural Networks? conference February 2017
The Power-optimised Software Envelope
  • Roberts, Stephen I.; Wright, Steven A.; Fahmy, Suhaib A.
  • ACM Transactions on Architecture and Code Optimization, Vol. 16, Issue 3 https://doi.org/10.1145/3321551
journal June 2019
Flexible Communication Avoiding Matrix Multiplication on FPGA with High-Level Synthesis
  • de Fine Licht, Johannes; Kwasniewski, Grzegorz; Hoefler, Torsten
  • Proceedings of the 2020 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays https://doi.org/10.1145/3373087.3375296
conference February 2020
A CAD-based methodology to optimize HLS code via the roofline model conference November 2020
Performance Modeling for FPGAs: Extending the Roofline Model with High-Level Synthesis Tools journal January 2013

Similar Records

Related Subjects

97 MATHEMATICS AND COMPUTING