Automated processor specification and task allocation for embedded multicomputer systems: The packing-based approaches

Abstract

This paper considers the coupled design problems of processor specification and task allocation for embedded multicomputer systems. A packing-based representation is proposed that allows the problems to be solved concurrently. An algorithm based on this representation is described that utilizes a new heuristic packing technique coupled with an incremental design advisor. This algo rithm, named [DAT, was benchmarked against three base line algorithms on a combination of real and synthetic test cases with respect [0 two figures of merit: hardware cost and run-rime. The real TeST cases are based on commer cially developed automotive electronic applications and the baseline algorithms represent a mixture of search. heu ristic and simulated annealing approaches. For all test cases, the IDAT algorithm was found to generate near optimal solutions with up to three orders of magnitude improvement in run-time compared to the baseline algo rithms.

The design of mixed hardware/software systems

“…The goal is to meet some performance objective while minimizing the cost of the hardware, either by choosing inexpensive processing elements, reducing the number of processing elements, or both. Several approaches to the design of heterogeneous multiprocessing systems have been published recently [9] [12] [13]. The key difference in the approaches is the way the processing elements are specified and the way the optimization is performed.…”

Modeling multicomputer task allocation as a vector packing problem

“…It serves as an abstraction for evaluating task allocation alternatives. For more detailed information, refer to [4].…”

“…[2,28]). The constraints used during experimentation are not shown here, but details can be found in [4].…”

“…In this situation, since there is no restriction on packing the scalar bin, it can be ignored and the problem reduces to the widely investigated vector packing problem. Furthermore, the packing problem shown in Figure 2 is NPcomplete, since vector packing is NP-complete and the reduction described above is clearly a polynomial transformation [4].…”

“…The reason for this is also intuitive: There is no notion of what constitutes a "best fit" for a vector object being placed into a vector bin. In fact, the bin that is the "best fit" for any particular dimension may actually be a poor choice across the remaining dimensions [4]. Selecting the vector bin that minimizes utilization of the scalar bin (i.e.…”