본문 바로가기

Roofline-Model-Based Design Space Exploration for Dataflow Techniques of CNN Accelerators

문서 유형
연구출판물 · 학술지 논문
등재
SCIEScopus
저널 순위
JCR Q1Scopus Q1
저널
IEEE ACCESS8 · pp.172509 · 2020
피인용
Web of Science 19Scopus 19
FWCI (Scopus)
0
JCR IF
JCR 2019 3.745
ISSN
2169-3536

저자

  1. 박찬제1저자Department of Electronics Engineering, Pusan National University, Pusan 46241, South Korea
  2. 박성경Department of Electronics Engineering, Pusan National University, Pusan 46241, South Korea
  3. Park, Chester Sungchung교신저자Department of Electrical Engineering, Konkuk University, Seoul 05029, South Korea

초록

To effectively compute convolutional layers, a complex design space must exist (e.g., the dataflow techniques associated with the layer parameters, loop transformation techniques, and hardware parameters). For efficient design space exploration (DSE) of various dataflow techniques, namely, the weight-stationary (WS), output-stationary (OS), row-stationary (RS), and no local reuse (NLR) techniques, the processing element (PE) structure and computational pattern of each dataflow technique are analyzed. Various performance metrics are calculated, namely, the throughput (in giga-operations per second, GOPS), computation-to-communication ratio (CCR), on-chip memory usage, and off-chip memory bandwidth, as closed-form expressions of the layer and hardware parameters. In addition, loop interchange and loop unrolling techniques with a double-buffer architecture are assumed. Many roofline model-based simulations are performed to explore relevant dataflow techniques for a wide variety of convolutional layers of typical neural networks. Through simulation, this paper provides insights into the trends in accelerator performance as the layer parameters change. For convolutional layers with large input and output feature map (ifmap and ofmap) widths and heights, the GOPS of the NLR dataflow technique tends to be higher than that of the techniques. For convolutional layers with low weight and ofmap widths and heights, the RS dataflow technique achieves optimal GOPS and on-chip memory usage. In the case of convolutional layers with small weight widths and heights, the GOPS of the WS dataflow technique tends to be high. In the case of convolutional layers with small ofmap widths and heights, the OS dataflow technique achieves optimal GOPS and on-chip memory usage.

키워드

Computational modelingBandwidthSpace explorationHardwareConvolutionMemory managementAcceleratorconvolutional neural networks (CNNs)dataflow techniquesrooflinesimulationprocessing element (PE)design space exploration (DSE)field-programmable gate array (FPGA)

식별자

Scopus EID
2-s2.0-85102776061
WoS UID
000575884000001