Abstract: A cache memory system including a cache memory having a plurality of cache lines. An index portion of a tag array includes an n-bit pointer entry for every cache line. A shared tag portion of a tag array includes a number of entries, where each entry includes shared tag information that is shared among a plurality of the cache lines. Each n-bit pointer in the index portion of the tag array points into an entry in the shared tag portion.

Abstract: A cache system including a data cache memory comprising a plurality of cache lines. A tag store has an entry representing each line in the cache memory where each entry comprises tag information for accessing the data cache. The tag information includes state information indicating whether the represented cache line includes dirty data. A speculative write back unit monitors the state information and is operative to initiate a write back of a cache line having more than a preselected amount of dirty data.

Abstract: A system and method for controlling data transmission between two network elements. A first port of a transmitting element is coupled to a second port of a receiving element. The second port includes buffers for temporarily storing received data until the data can be sent to another element. Included in the transmitting element are a received-currently-full register (RCFR), a sent-and-not-received register (SANRR), and a buffer-busy register (BBR). The transmitting element checks its BBR to determine if a buffer in the receiving element is available. The availability of buffers can be determined using a single priority protocol or a multiple priority protocol.

Abstract: A method and apparatus for detecting and isolating an interconnect fault in a packet switched network generates a parity check error code for status messages which are used for flow control in a packet switched network. The packet switched network uses a reverse flow control method wherein status messages are sent locally between adjacent nodes. A receiving node uses status messages to inform an adjacent node of the availability of the input buffers located in the receiving node. Included in the status message is a parity check code that is sent sequentially with the status message using two phases of a clock. The parity check code is a one bit parity check for each bit of the status message. Faults on the local interconnect are detected at the receiving node by performing a one bit parity check on the received status message using the accompanying parity code.

Abstract: A multiprocessor system includes a plurality of nodes and an interconnect that includes routers. Each node includes a reliable packet mover and a fast frame mover. The reliable packet mover provides packets to the fast frame mover which adds routing information to the packet to form a frame. The route to each node is predetermined. The frame is provided to the routers which delete the route from the routing information. If the frame is lost while being routed, the router discards the frame. If the packet is received at a destination node, the reliable packet mover in that node sends an acknowledgment to the source node if the packet passes an error detection test. The reliable packet mover in the source node resends the packet if it does not receive an acknowledgment in a predetermined time. The fast frame mover randomly selects the route from a plurality of predetermined routes to the destination node according to a probability distribution.

Abstract: An plesioasynchronous and asynchronous router circuit communicates with neighboring router circuits and nodes. Each of the router circuits includes a plurality of input ports for receiving frames of data and a plurality of output ports for transmitting frames of data. Each router circuit further includes a plurality of input buffers for storing frames of data received at an input port, and an arbiter system for choosing one of several input buffers associated with a particular one of said output ports. The arbiter system includes a plurality of arbiter subsystems associated with corresponding ones of said plurality of output ports. The plesioasynchronous and asynchronous router circuit further includes a crossbar switch for connecting an arbiter selected input buffer with a particular one of said output ports.

Abstract: A method and apparatus for detecting and isolating an interconnect fault in a packet switched network generates a parity check error code for status messages which are used for flow control in a packet switched network. The packet switched network uses a reverse flow control method wherein status messages are sent locally between adjacent nodes. A receiving node uses status messages to inform an adjacent node of the availability of the input buffers located in the receiving node. Included in the status message is a parity check code that is sent sequentially with the status message using two phases of a clock. The parity check code is a one bit parity check for each bit of the status message. Faults on the local interconnect are detected at the receiving node by performing a one bit parity check on the received status message using the accompanying parity code.

Abstract: A system and method dynamically determines the topology of a source node routing network while having a minimal effect on network performance and without requiring expensive hardware to implement. A source node generates a ping frame. The source node transmits the ping frame to a first source router that is coupled to the source node. The first router transparently identifies the frame as a ping frame and creates an echo frame that is transmitted back to the source node. The first router identifies the port from which the ping frame is received and places this information in the header of the echo frame along with an echo frame identifier. The source node receives the echo frame and identifies routers and nodes to which a ping frame has not been sent based upon the connectivity information in the received echo frame. The source node continue generating and transmitting ping frame to all nodes and routers in the network.

Abstract: A system and method for transferring data between clock domains operating at substantially the same frequency continuously compares the phase relationship of the clocks of the two domains and retains the current state of comparison at the start of a transfer of a block or frame of data for determining along which one of multiple data paths within the synchronizing circuit the transfer of the data frame will take place. Several data paths with different delays (at least two) transfer the data frame and clock signals. A phase comparator responds to the phase relationship between clocks attaining a value within one or another range of values at the start of a data frame to determine which one of the multiple data paths transfers the data frame.