[Cuda] Not yet support ewise after reduce


I’m trying to compile tensorflow resnet_v1_50 model for intel graphics target. I got an error on this line: https://github.com/dmlc/tvm/blob/master/topi/python/topi/cuda/reduction.py#L110 A possible reason might be tf model uses broadcast_add to represent elemwise_add, and some broadcast_add follow after reduction ops. Any solution for this issue?