Guinness - GUI based Neural Network Synthesizer



Guinness is a GUI based framework that includes both a training on a GPU, and a bitstream generation for an FPGA using the Xilinx SDSoC. This tool uses the Chainer deep learning framework to train a binarized CNN. Also, it uses optimization techniques for an FPGA implementation.

Compared with the conventional FPGA realizations, although the classification accuracy is 6.5% decayed, the performance is 2.45 times faster, the power efficiency is slightly better, and the area efficiency is 2.68 times better. Compared with the ARM Cortex-A57, it is 136.8 times faster, it dissipates 3.1 times much power, and its performance per power efficiency is 44.7 times better. Also, compared with the Maxwell embedded GPU, it is 4.9 times faster, it dissipates 1.3 times much power, and its performance per power efficiency is 3.8 times better. 

Details are shown in following papers:

[Nakahara IPDPSW2017] H. Yonekawa and H. Nakahara, "On-Chip Memory Based Binarized Convolutional Deep Neural Network Applying Batch Normalization Free Technique on an FPGA," IPDPS Workshops, 2017, pp. 98-105.
[Nakahara FPL2017] H. Nakahara et al., "A Fully Connected Layer Elimination for a Binarized Convolutional Neural Network on an FPGA", FPL, 2017, (to appear).


Comments

Popular posts from this blog

Xilinx AXI Stream tutorial - Part 1

Analysis, elaboration and synthesis