A Network Service for DSP Multicomputers
ICA3PP '08 Proceedings of the 8th international conference on Algorithms and Architectures for Parallel Processing
Hi-index | 0.00 |
Video, image and signal processing applications show a high computational complexity and real-time restrictions. Hence they demand task distribution on DSP multi-computers. Unfortunately, DSP vendors offer these platforms nowadays with proprietary kernels and hardware specific communication libraries. As a result, current applications are hardware targeted and not portable at all. We understand that the use of MPI in the signal processing world should help to improve this scenery. This article describes a preliminary implementation of MPI for Sundance machines. Size and performance figures are given.