We have detected your current browser version is not the latest one. Xilinx.com uses the latest web technologies to bring you the best online experience possible. Please upgrade to a Xilinx.com supported browser:Chrome, Firefox, Internet Explorer 11, Safari. Thank you!

Showing results for 
Search instead for 
Did you mean: 

DeePhi’s Zynq-based CNN processor is faster, more energy efficient than CPUs or GPUs

Xilinx Employee
Xilinx Employee
0 0 54.7K


Step right up folks, the Hot Chips show is unfolding this week at the Flint Center for the Performing Arts in Cupertino, California and DeePhi has rolled out a convolutional neural network (CNN) acceleration processor named Aristotle that is faster and more energy efficient than CPUs or GPUs performing the same tasks. Is DeePhi’s CNN Processor some sort of super ASIC based on carbon nanotubes or memristors and powered by Mr. Fusion? Nope.


DeePhi’s Aristotle CNN Processor is based on a Xilinx Zynq-7000 All Programmable SoC.





DeePhi Aristotle Processor Slide from Hot Chips 2016



According to this article on Nextplatform.com, DeePhi’s CEO and co-founder Song Yao said that “CPUs don’t have the energy efficiency, GPUs are great for training but lack ‘the efficiency in inference’, DSP approaches don’t have high enough performance and have a high cache miss rate and of course, ASICs are too slow to market—and even when produced, finding a large enough market to justify development cost is difficult.


“FPGA based deep learning accelerators meet most requirements. They have acceptable power and performance, they can support customized architecture and have high on-chip memory bandwidth and are very reliable.”


Yao reportedly said: “The FPGA based DPU platform achieves an order of magnitude higher energy efficiency over GPU on image recognition and speech detection.” He also said that “traditionally, FPGA based acceleration via hand coding took a few months; using OpenCL and the related tool chain brings that down to one month.”


DeePhi was launched in March, 2016 and leverages the results of work from teams at Stanford University and Tsinghua University.




Tags (2)