What FPGA accelerator algorithm for CNN's have significant reduction in resource usage while maintaining high accuracy and performance comparable to normal algorithm.
I'm training the network in a PC and planning to export the inference into a FPGA.
One of the suggestion will be to convert floating point implementation to fix point or integer. Not sure how much it will affect the accuracy. This usually depends on what the data set , weights and layers are.
Xilinx have DPU IPs. These are soft IPs, so does use resources inside the FPGA.
Usually the bottle neck is data transfer from memory into FPGAs. One way to get out of this will be to come up with smart BRAM Architecture to store part of data/weights.
Article Design of FPGA-Based Accelerator for Convolutional Neural Ne...
CPU has insufficient resources to satisfy the efficient computation of the convolution neural network (CNN), especially for embedded applications. Therefore, heterogeneous computing platforms are widely used to accelerate CNN tasks, such as GPU, FPGA, and ASIC. Among these, FPGA can accelerate the computation by mapping the algorithm to the parallel hardware instead of CPU, which cannot fully exploit the parallelism. By fully using the parallelism of the neural network’s structure, FPGA can reduce the computing costs and increase the computing speed. However, the development of FPGA requires great design skills. As a heterogeneous development platform, OpenCL has some advantages such as high abstraction level, short development cycle, and strong portability, which can make up for the lack of skilled designers. This paper uses Xilinx SDAccel to realize the parallel acceleration of CNN task, and it also proposes an optimizing strategy of single convolutional layer to accelerate CNN. Simulation results show that the calculation speed could be improved by adopting the proposed optimizing strategy.
Chandler Timm Cagmat Doloriel It depends on the architecture of the neural network and also the type of application that you are targeting. Which FPGA device are you using?
- Badges
- Science topic
- Similar topics
- Philosophy
- Philosophy Of Science
- Reasoning