关键词: 布局优化, 可布线性预测, ResNet50, 可编程逻辑阵列

Abstract: At present, the mainstream FPGA design scale in industry has exceeded 100 million gate level, and the traditional placement algorithm optimization has been unable to meet the design needs of users for ultra-large scale circuits because the optimization objectives can not be completely consistent with the actual routing requirements .To solve this problem, this study proposes a placement method based on the improved ResNet50, establishes a data mapping model between the place parameters and the rout ability. At the same time a collaborative optimization method before and after place and route is proposed,that is the placement parameters is used as the training input of AI model, the routing result is used as the training target to train AI model. And the trained model is used to optimize the place. According to the testing result, this method can be proved effectively improved the rout ability.

Key words: placement optimization, routeability prediction, ResNet50, FPGA