The Warp project is a study in high-performance, programmable systolic architectures. Although the feasibility of the concept of systolic computing has long been established, the results have mainly been theoretical in nature, and many lower-level architectural and implementation issues have not been studied and addressed. We believed that these practical issues would not even be exposed, let alone resolved, without implementation and experimentation. We did not just set out to implement a systolic array to perform a specific function, we extended the concept of systolic processing by implementing each processing unit as a programmable, high-performance processor. The tenet of the project is that the theories and results in systolic computing research are applicable also to arrays of processors that are programmable and powerful.