CU.POKer: Placing DNNs on Wafer-Scale Al Accelerator with Optimal Kernel Sizing | IEEE Conference Publication | IEEE Xplore