1. Automatically generate Verilog code tools based on the hardware architecture of convolutional neural networks: 4 different hardware architectures (output stationary, weight stationary, Tree architecture, NVDLA) can be generated for the currently more commonly used DNN networks.
2. Visual performance index analysis tool: According to the selected DNN model, and the choice of hardware architecture specifications, analyze the performance index.
1. Automatically generate Verilog code tool based on the hardware architecture of convolutional neural network: Add the supported layer type: layer type: shortcut, route, upsample, depthwise convolution.
2. Visual performance index analysis tool: It can effectively analyze the performance of the model on the hardware architecture, and select the appropriate hardware specifications to achieve the hardware acceleration of the DNN model.
Nowadays, many DNN models are used in various applications, and many of them require hardware acceleration. To design the hardware accelerator architecture, different algorithm and architecture options should be considered. For this reason, we design a general-purpose automatic generation of DNN hardware accelerator for the current popular DNN.
We also design the Profiler to save the time of testing performance. We have a function for quick analysis of various performance indicators, so that users can select the most suitable hardware architecture for the model in the shortest time.

線上展網址：
https://tievirtual.twtm.com.tw/iframe/0e63b016-6c1c-4957-9715-b0386dda2fe4?group=23bfb1fa-dd5b-4836-81a1-4a1809b1bae5&lang=en