Voltage Scaled Low Power DNN Accelerator Design on Reconfigurable Platform
-
Published:2024-04-10
Issue:8
Volume:13
Page:1431
-
ISSN:2079-9292
-
Container-title:Electronics
-
language:en
-
Short-container-title:Electronics
Author:
Paul Rourab12, Sarkar Sreetama3ORCID, Sau Suman4, Roy Sanghamitra5, Chakraborty Koushik5, Chakrabarti Amlan6ORCID
Affiliation:
1. Computer Science & Engineering, Siksha O Anusandhan, Bhubaneswar 751030, India 2. Computer Science, University of Pisa, 56127 Pisa, Italy 3. Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA 90089, USA 4. Computer Science & Information Technology, Siksha O Anusandhan, Bhubaneswar 751030, India 5. Department Electrical and Computer Engineering, Utah State University, Logan, UT 84322, USA 6. School of IT, University of Calcutta, Kolkata 700019, India
Abstract
The exponential emergence of Field-Programmable Gate Arrays (FPGAs) has accelerated research on hardware implementation of Deep Neural Networks (DNNs). Among all DNN processors, domain-specific architectures such as Google’s Tensor Processor Unit (TPU) have outperformed conventional GPUs (Graphics Processing Units) and CPUs (Central Processing Units). However, implementing low-power TPUs in reconfigurable hardware remains a challenge in this field. Voltage scaling, a popular approach for energy savings, can be challenging in FPGAs, as it may lead to timing failures if not implemented appropriately. This work presents an ultra-low-power FPGA implementation of a TPU for edge applications. We divide the systolic array of a TPU into different FPGA partitions based on the minimum slack value of different design paths of Multiplier Accumulators (MACs). Each partition uses different near-threshold (NTC) biasing voltages to run its FPGA cores. The biasing voltage for each partition is roughly calculated by the proposed static schemes. However, further calibration of biasing voltage is performed by the proposed runtime scheme. To overcome the timing failure caused by NTC, the MACs with higher minimum slack are placed in lower-voltage partitions, while the MACs with lower minimum slack paths are placed in higher-voltage partitions. The proposed architecture is implemented in a commercial platform, namely Vivado with Xilinx Artix-7 FPGA and academic platform VTR with 22 nm, 45 nm and 130 nm FPGAs. Any timing error caused by NTC can be caught by the Razor flipflop used in each MAC. The proposed voltage-scaled, partitioned systolic array can save 3.1% to 11.6% of dynamic power in Vivado and VTR tools, respectively, depending on the FPGA technology, partition size, number of partitions and biasing voltages. The normalized performance and accuracy of benchmark models running on our low-power TPU are very competitive compared to existing literature.
Funder
National Science Foundation
Reference28 articles.
1. Kimm, H., Paik, I., and Kimm, H. (2021, January 20–23). Performance Comparision of TPU, GPU, CPU on Google Colaboratory Over Distributed Deep Learning. Proceedings of the 2021 IEEE 14th International Symposium on Embedded Multicore/Many-Core Systems-on-Chip (MCSoC), Singapore. 2. Caulfield, A.M., Chung, E.S., Putnam, A., Angepat, H., Fowers, J., Haselman, M., Heil, S., Humphrey, M., Kaur, P., and Kim, J.Y. (2016, January 15–19). A cloud-scale acceleration architecture. Proceedings of the 2016 49th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO), Taipei, Taiwan. 3. A Reconfigurable Fabric for Accelerating Large-Scale Datacenter Services;Putnam;IEEE Micro,2015 4. Salami, B., Onural, E.B., Yuksel, I.E., Koc, F., Ergin, O., Cristal Kestelman, A., Unsal, O., Sarbazi-Azad, H., and Mutlu, O. (July, January 29). An Experimental Study of Reduced-Voltage Operation in Modern FPGAs for Neural Network Acceleration. Proceedings of the 2020 50th Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN), Valencia, Spain. 5. GreenTPU: Predictive Design Paradigm for Improving Timing Error Resilience of a Near-Threshold Tensor Processing Unit;Pandey;IEEE Trans. Very Large Scale Integr. (VLSI) Syst.,2020
|
|