Albert Van Der Werf

PHIDEO: High-level synthesis for high throughput applications

This paper describes a new approach to high-level synthesis for high throughput applications. Such applications are typically found in real-time video systems such as HDTV. The method is capable of dealing with hierarchical flow graphs containing loops with manifest boundaries and linear index expressions. The algorithm is based on the model of periodic operations which allows optimizations across loop boundaries. Processing units and storage units are minimized simultaneously. The algorithm is implemented in thePHIDEO system. The major parts of this system are the processing unit synthesis, the scheduler and the memory synthesis including address generation.

Mapping array communication onto FIFO communication - towards an implementation

In high-throughput real-time media processing systems, the communication between processing units is typically specified as multi-dimensional arrays. However, the implementation of such applications is mostly FIFO-based. Mapping array communication onto a FIFO-based implementation requires complex address generators if the arrays have multiple dimensions. In this paper, we present a method for mapping array communication onto an efficient microcomputer architecture implementation based on FIFO communication via shared memory. A good hardware/software partitioning for the address generation is proposed. Furthermore, a complete design flow from specification to implementation is described. We illustrate this method with a design case: the communication of video frames between the frontend and the compressor in an MPEG encoder.

Multidimensional Periodic Scheduling Model and Complexity

We discuss the multidimensional periodic scheduling problem, which originates from the design of high-throughput real-time digital signal processing systems. We introduce the concept of multidimensional periodic operations in order to cope with problems originating from loop hierarchies and explicit timing requirements. We present a model of the multidimensional periodic scheduling problem and show that this problem and two related sub-problems are NP-hard. Furthermore, we identify several special cases induced by practical situations. Some of these special cases are proven to be well-solvable. Finally, we present a sketch of a solution approach.