Innovus somehow detects FF synchronizers in the circuit (which is good) and disables InPlaceOptimization of synchronizer nets. However, it seems that negative hold time slacks of non-optimized nodes prevent to meet the desired hold time slack in the rest of the circuit. My project is based on 180 nm technology and I target 0.2 ns hold time slack. In the presence of synchronizer nets with ~ -0.2 ns hold time slack, the best positive hold time slack after 'opt_design -post_route -setup -hold' is reported in the range of 0.03 ns.
Are there any suggestions how to implement circuits with synchronizers ?
Is there any way to exclude disabled nets from the optimization ?
dusan.raic@fe.uni-lj.si
or
raic.dusan@gmail.com