DiMO-CNN: Deep Learning Toolkit-Accelerated Analytical Modeling and Optimization of CNN Hardware and Dataflow

The growing complexity of CNNs demands both hardware acceleration design and dataflow mapping solutions. The large co-design solution space presents a huge challenge. We introduce an analytical model for assessing CNN hardware design and dataflow solutions, using a matrix-based approach. Our co-optimization method, combining nonlinear programming and parallel local search, excels in addressing the power-performance-area tradeoff. The average relative error of our analytical model compared with Timeloop is as small as 1%.Compared to state-of-the-art methods, our co-optimization achieves solutions with average 3.14× shorter inference latency, 68.2% less power consumption, and 74% less area on all testcases. It also provides a 200× speedup of optimization runtime.