Maspar Computer Corporation

W Thomas Blank

John R Nickolls

Won Sub Kim

John Zapisek

Kevin Van Horn

Jeffrey C Kalb

David M Bulfer

Jeffery C Kalb

Hannes Figel

A massively parallel computer system is disclosed having a global router network in which pipeline registers are spatially distributed to increase the messaging speed of the global router network. The global router network includes an expansion tap for processor to I/O messaging so that I/O messaging bandwidth matches interprocessor messaging bandwidth. A route-opening message packet includes protocol bits which are treated homogeneously with steering bits. The route-opening packet further includes redundant address bits for imparting a multiple-crossbars personality to router chips within the global router network. A structure and method for spatially supporting the processors of the massively parallel system and the global router network are also disclosed.

Scalable processor to processor and processor to I/O interconnection

A massively parallel processor is provided with a plurality of clusters. Each cluster includes a plurality of processor elements ('PEs') and a cluster memory. Each PE of the cluster has associated with it an address register, a stage register, an error register, a PE enable flag, a memory flag, and a grant request flag. A cluster data bus and an error bus connects each of the stage registers and error registers of the cluster to the memory. The grant request flags of the cluster are interconnected by a polling network, which polls only one of the grant request flags at a time. In response to a signal on the polling network and the state of the associated memory flag, the grant request flag determines an I/O operation between the associated data register and the cluster memory over the cluster data bus. Both data and error bits are associated with respective processor elements. The sequential memory operations proceed in parallel with parallel processor operations. The sequential memory operations also may be queued. Addressing modes include direct and indirect. In direct address mode, a PE addresses its own address space by appending its PE number to a broadcast partial address. The broadcast partial address is furnished over a broadcast bus, and the PE number is furnished on a cluster address bus. In indirect address mode, a PE addresses either its own address space or that of other PEs in its cluster by locally calculating a partial address, then appending to it either its own PE number or that of another PE in its cluster. The full address is furnished over the cluster address bus.

Parallel processor memory transfer system using parallel transfers

Scalable processor to processor and processor-to-I/O interconnection

Growing community of inventors

Palo Alto, CA, United States of America

W Thomas Blank