Executive Summary: Computational Dataflow is a program behavior that requires a relatively easy-to-implement hardware organization. The behavior formally defines the flow of values between different operations in a program. Historically, when dataflow has been implemented as a hardware architecture the resulting systems have invariably fallen short of their promise to deliver high-performance. We observe that these programs almost always intertwine the dataflow behavior with program change-in-control behavior, and that this combination has hobbled past efforts. We assert that the simple, yet novel idea of embedding control into a dataflow chip’s organization results in improvements that surpass the long-held performance and efficiency barriers.