Abstract: The cache coherency protocol described herein can be used to maintain a virtual model of a system, where the virtual model does not change as the system configuration changes. In general, the virtual model is based on the assumption that each node in the system can directly communicate with some number of other nodes in the system. In one embodiment, for each cache line, the address of the cache line is used to designate a node as the “home” node and all other nodes as “peer” nodes. The protocol specifies one set of messages for communication with the line's home node and another set of messages for communication with the line's peer nodes.

Abstract: A system and method for reducing linear address aliasing is described. In one embodiment, a portion of a linear address is combined with a process identifier, e.g., a page directory base pointer to form an adjusted-linear address. The page directory base pointer is unique to a process and combining it with a portion of the linear address produces an adjusted-linear address that provides a high probability of no aliasing. A portion of the adjusted-linear address is used to search an adjusted-linear-addressed cache memory for a data block specified by the linear address. If the data block does not reside in the adjusted-linear-addressed cache memory, then a replacement policy selects one of the cache lines in the adjusted-linear-addressed cache memory and replaces the data block of the selected cache line with a data block located at a physical address produced from translating the linear address.

Abstract: A conflict resolution technique provides consistency such that all conflicts can be detected by at least one of the conflicting requestors if each node monitors all requests after that node has made its own request. If a line is in the Exclusive, Modified or Forward state, conflicts are resolved at the node holding the unique copy. The winner of the conflict resolution, and possibly the losers, report the conflict to the home node, which pairs conflict reports and issues forwarding instructions to assure that all requesting nodes eventually receive the requested data. If a requested cache line is either uncached or present only in the Shared state, the home node provides a copy of the cache node and resolves conflicts. In one embodiment, a blackout period after all responses until an acknowledgement message has been received allows all conflicting nodes to be aware of conflicts in which they are involved.

Abstract: Described herein is a cache coherency protocol having five states: Modified, Exclusive, Shared, Invalid and Forward (MESIF). The MESIF cache coherency protocol includes a Forward (F) state that designates a single copy of data from which further copies can be made. A cache line in the F state is used to respond to request for a copy of the cache line. In one embodiment, the newly created copy is placed in the F state and the cache line previously in the F state is put in the Shared (S) state, or the Invalid (I) state. Thus, if the cache line is shared, one shared copy is in the F state and the remaining copies of the cache line are in the S state.

Abstract: Described herein is a cache coherency protocol having five states: Modified, Exclusive, Shared, Invalid and Forward (MESIF). The MESIF cache coherency protocol includes a Forward (F) state that designates a single copy of data from which further copies can be made. A cache line in the F state is used to respond to request for a copy of the cache line. In one embodiment, the newly created copy is placed in the F state and the cache line previously in the F state is put in the Shared (S) state, or the Invalid (I) state. Thus, if the cache line is shared, one shared copy is in the F state and the remaining copies of the cache line are in the S state.

Abstract: Use of an import cache and/or an export directory with an agent within to respond to requests for data. The import cache stores data that has been imported through the agent. The export directory stores information related to data that has been exported through the agent. Because the import cache and the export directory only store data that has passed through the agent, not all data transferred within a system are tracked by a single import cache or export directory.

Abstract: A conflict resolution technique provides consistency such that all conflicts can be detected by at least one of the conflicting requesters if each node monitors all requests after that node has made its own request. If a line is in the Exclusive, Modified or Forward state, conflicts are resolved at the node holding the unique copy. The winner of the conflict resolution, and possibly the losers, report the conflict to the home node, which pairs conflict reports and issues forwarding instructions to assure that all requesting nodes eventually receive the requested data. If a requested cache line is either uncached or present only in the Shared state, the home node provides a copy of the cache node and resolves conflicts. In one embodiment, a blackout period after all responses until an acknowledgement message has been received allows all conflicting nodes to be aware of conflicts in which they are involved.

Abstract: A conflict resolution technique provides consistency such that all conflicts can be detected by at least one of the conflicting requestors if each node monitors all requests after that node has made its own request. If a line is in the Exclusive, Modified or Forward state, conflicts are resolved at the node holding the unique copy. The winner of the conflict resolution, and possibly the losers, report the conflict to the home node, which pairs conflict reports and issues forwarding instructions to assure that all requesting nodes eventually receive the requested data. If a requested cache line is either uncached or present only in the Shared state, the home node provides a copy of the cache node and resolves conflicts. In one embodiment, a blackout period after all responses until an acknowledgement message has been received allows all conflicting nodes to be aware of conflicts in which they are involved.

Abstract: The cache coherency protocol described herein can be used to maintain a virtual model of a system, where the virtual model does not change as the system configuration changes. In general, the virtual model is based on the assumption that each node in the system can directly communicate with some number of other nodes in the system. In one embodiment, for each cache line, the address of the cache line is used to designate a node as the “home” node and all other nodes as “peer” nodes. The protocol specifies one set of messages for communication with the line's home node and another set of messages for communication with the line's peer nodes.

Abstract: A system and method for storing only one copy of a data block that is shared by two or more processes is described. In one embodiment, a global/non-global predictor predicts whether a data block, specified by a linear address, is shared or not shared by two or more processes. If the data block is predicted to be non-shared, then a portion of the linear address referencing the data block is combined with a process identifier that is unique to form a global/non-global linear address. If the data block is predicted to be shared, then the global/non-global linear address is the linear address itself. If the prediction as to whether or not the data block is shared is incorrect, then the actual value of whether or not the data block is shared is used in computing a corrected global/non-global linear address.

Abstract: An embodiment of the present invention provides a memory controller that includes a plurality of transaction queues and an arbiter, a prefetch cache in communication with the arbiter, and a prefetch queue in communication with the prefetch cache. The prefetch queue also may be provided in communication with each of the transaction queues for the purpose of determining whether the transaction queues are operating in a congested state.

Abstract: A system and method for storing only one copy of a data block that is shared by two or more processes is described. In one embodiment, a global/non-global predictor predicts whether a data block, specified by a linear address, is shared or not shared by two or more processes. If the data block is predicted to be non-shared, then a portion of the linear address referencing the data block is combined with a process identifier that is unique to form a global/non-global linear address. If the data block is predicted to be shared, then the global/non-global linear address is the linear address itself. If the prediction as to whether or not the data block is shared is incorrect, then the actual value of whether or not the data block is shared is used in computing a corrected global/non-global linear address.

Abstract: A system and method for storing only one copy of a data block that is shared by two or more processes is described. In one embodiment, a global/non-global predictor predicts whether a data block, specified by a linear address, is shared or not shared by two or more processes. If the data block is predicted to be non-shared, then a portion of the linear address referencing the data block is combined with a process identifier that is unique to form a global/non-global linear address. If the data block is predicted to be shared, then the global/non-global linear address is the linear address itself. If the prediction as to whether or not the data block is shared is incorrect, then the actual value of whether or not the data block is shared is used in computing a corrected global/non-global linear address.

Abstract: A hybrid branch predictor is disclosed. The predictor includes prediction aiding information, a plurality of branch predictors to provide a plurality of branch predictions, a plurality of storage elements to hold less than full extent of the branch predictions, but sharing information among said plurality of storage elements enables extraction of said full extent of the prediction. The predictor also includes a selection mechanism to select a prediction from the plurality of branch predictions.

Abstract: A system and method for reducing linear address aliasing is described. In one embodiment, a portion of a linear address is combined with a process identifier, e.g., a page directory base pointer to form an adjusted-linear address. The page directory base pointer is unique to a process and combining it with a portion of the linear address produces an adjusted-linear address that provides a high probability of no aliasing. A portion of the adjusted-linear address is used to search an adjusted-linear-addressed cache memory for a data block specified by the linear address. If the data block does not reside in the adjusted-linear-addressed cache memory, then a replacement policy selects one of the cache lines in the adjusted-linear-addressed cache memory and replaces the data block of the selected cache line with a data block located at a physical address produced from translating the linear address.

Abstract: A system and method for storing only one copy of a data block that is shared by two or more processes is described. In one embodiment, a global/non-global predictor predicts whether a data block, specified by a linear address, is shared or not shared by two or more processes. If the data block is predicted to be non-shared, then a portion of the linear address referencing the data block is combined with a process identifier that is unique to form a global/non-global linear address. If the data block is predicted to be shared, then the global/non-global linear address is the linear address itself. If the prediction as to whether or not the data block is shared is incorrect, then the actual value of whether or not the data block is shared is used in computing a corrected global/non-global linear address.

Abstract: A technique for providing a dirty line cache to supplement a cache memory, in order to improve caching performance for a processor in a computer system. A fully-associative cache memory is coupled to operate with a main cache memory at a particular level of the cache hierarchy. The supplemental cache is termed as a dirty line cache, since it only stores dirty cache lines. In the preferred embodiment, the dirty line cache is implemented in a write-out buffer of a write-back or write-through cache.