期刊名称：Oriental Journal of Computer Science and Technology
印刷版ISSN：0974-6471
出版年度：2013
卷号：6
期号：2
页码：09-14
语种：English
出版社：Oriental Scientific Publishing Company
摘要：ANnSP is a stream-based programmable and code-level statically reconûgurable processor for realization of neural networks in embedded systems. ANnSP is provided with a neural-network-to-stream compiler and a hardware core builder. The ANnSP stream compiler makes it possible to realize various neural networks using ANnSP. On the other hand, the ANnSP builder makes the ANnSP processor an IP core that can be restructured to satisfy different demands and constraints. This paper presents the architecture of the ANnSP processor, the streaming mechanism, and the builder facilities. Also, synthesis results of a 64-PE ANnSP on a 0.18µm standard-cell library are presented. The obtained results show that a 64-PE ANnSP can perform computations of 25.6 giga connections in a second, while its throughput is upto 51.2 giga 32-bit ûxed point operations per second. Comparing with high performance parallel architectures locates 64-PE ANnSP among the best state of the art parallel processors.
Loading...
