초록 ▼

A systolic array system of inner product step processors is provided in the form of a mesh connected network which rhythmically compute and pass data through the system. Each processor in the system regularly feeds data in and out, each time performing some computation, so that a regular flow of dat

A systolic array system of inner product step processors is provided in the form of a mesh connected network which rhythmically compute and pass data through the system. Each processor in the system regularly feeds data in and out, each time performing some computation, so that a regular flow of data is kept up in the network. Many basic matrix computations can be readily and efficiently pipelined on systolic array network systems according to these inventions. Such arrays enjoy simple and regular communication paths and the individual processors in the networks are substantially all identical. Similar hexagonally connected processors can, for example, optionally perform matrix multiplication and LU-decomposition of a matrix. Linearly connected systolic arrays are useful for performing a variety of other computations.

대표청구항 ▼

In a computer system having external handling means delivering input to be processed and receiving output and a plurality of processors forming a network receiving said input, the improvement comprising said plurality of processors being independent of a central control which sends control instructi

In a computer system having external handling means delivering input to be processed and receiving output and a plurality of processors forming a network receiving said input, the improvement comprising said plurality of processors being independent of a central control which sends control instructions to all processors simultaneously and being made up of boundry processors lying on the boundry of the network and being one of inner product step processors and processors capable of performing usual arithmetic operations including division and internal processors within the area surrounded by said boundry processors and being inner product step processors, said processors being assembled in a network by nearest neighbor connections in which a regular flow of data is maintained, synchronizing means connected to each processor rythmically feeding data into and out of processors in said network and between said network and said external handling means and connections between at least one of the boundry processors and said external handling means whereby data is fed therebetween.