Exploitable results Novel parallel algorithms and new architectural methodologies Research has been carried out in order to develop new real time parallel algorithms and new very large scale integration (VLSI) architectures for a variety of applications, both in real time multidimensional signal processing and in numerical processing. Here, a clear need exists for novel parallel algebraic and numerical kernels and the combinations of kernels into complete algorithms. In order to allow future design of complex high throughput applications, novel technologies in the area of architectural synthesis techniques have been explored, which create a suitable architecture from a behavioural specification. The following major results have been achieved: a tool that converts nested loop algorithms written in the matlab environment to single assignment code and corresponding (reduced) dependence graph; an application specific processor for rendering of artificial scenes with photo-realism; publications dealing with high resolution signal processing algorithms, in particular related to direction, delay and/or frequency estimation in radar and communications; a machine independent environment for processing and visualizing 3D images; a set of general purpose load balancing strategies and algorithms including data locality preservation, as well as memory constraints handling; methodologies for efficient and portable parallel programming including support for specific redistribution of data and chaining of several computation kernals; new techniques to schedule and map loop nests; new tools implemented in prototype compilers; a novel systematic methodology for low power exploration at the system level of data transfer and storage architectures for data dominated multi media applications; a prototype tool, MMAlpha, for the synthesis of highly regular architectures from high level specifications. È in corso la ricerca di dati su OpenAIRE... Si è verificato un errore durante la ricerca dei dati su OpenAIRE Nessun risultato disponibile
