diff --git a/src/finn/custom_op/fpgadataflow/tlastmarker.py b/src/finn/custom_op/fpgadataflow/tlastmarker.py index 9a372179042a7236c15a02cf4191ec608836256c..3f3bf6e79c8ab6d485c5356a4c3bba3623220df2 100644 --- a/src/finn/custom_op/fpgadataflow/tlastmarker.py +++ b/src/finn/custom_op/fpgadataflow/tlastmarker.py @@ -17,8 +17,15 @@ class TLastMarker(HLSCustomOp): return my_attrs def execute_node(self, context, graph): - # TODO consider implementing rtlsim for TLastMarker - raise Exception("TLastMarker does yet not support execution") + # TLastMarker's behavior is only visible when doing + # rtlsim with stitched IP, since it marks the end + # of the current image/input sample. when executing + # inside FINN as a single node, this is not visible. + # so here we simply return the input as output + i_name = self.onnx_node.input[0] + o_name = self.onnx_node.output[0] + i_tensor = context[i_name] + context[o_name] = i_tensor def make_shape_compatible_op(self): # not supported for shape inference