This document gives a short overview of the MOVE project at the Delft University of Technology. The MOVE project concerns the automation of application-specific processor system design providing optimal cost-performance tradeoffs. The project is based on three tenets: RISC, instruction-level parallelism, and massive message-passing parallelism. RISC principles tell us to carefully quantify cost and performance of extra hardware functionality, taking also into account the design turnaround time, which is influenced by the design complexity. Instruction-level parallelism allows for high-performance computing nodes, whose functionality and performance (both processing and i/o performance) is adapted to specific applications or application domains. If the application permits and requires the power of massive parallelism, the nodes should be easily configurable to allow massive MIMD processing. The nodes have local memory, and communicate through the exchange of messages. Although this organization exploits the locality of computing available in many massive parallel applications, it does not preclude the implementation of (logical) shared memory.

The MOVE project finds its direct roots in the SCARCE project [6][5][4][3][2][1] and the design of communication processors for distributed computer systems [9][8][7]. In both projects flexible processor frameworks (for data processing and communication respectively) have been designed and implemented using the ASA Silicon compiler and NELSIS  VLSI design tools. The main goal of the project is to provide an efficient way for designing and building high-performance application-specific processor systems. These systems consist of one or more high-performance computational nodes. Within the project we distinguish two main threads: 1) a framework for the design of these computational nodes according to the application needs; 2) a framework for the design of distributed memory systems based on these nodes.

MOVE architectures are based on the transport-triggering concept (the MOVE concept for short). This concept and its main advantages are discussed next. Section gives a high level description of the MOVE framework, on which the current research is focussed. A project status overview concludes this short overview.