Directory-based prediction method and apparatus for shared-memory multiprocessor systems

• Main Authors: YOUNG, CLIFFORD REGINALD, KAXIRAS, STEFANOS
• Format: Patent
• Language: English; French; German
• Published: 12.12.2001
• Edition: 7
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

A set of predicted readers are determined for a data block subject to a write request in a shared-memory multiprocessor system by first determining a current set of readers of the data block, and then generating the set of predicted readers based on the current set of readers and at least one additional set of readers representative of at least a portion of a global history of a directory associated with the data block. In one possible implementation, the set of predicted readers are generated by applying a function to the current set of readers and one or more additional sets of readers. The function may be, for example, a union function, an intersection function or a pattern-based function, and the directory and data block may be elements of a memory associated with a particular processor node of the multiprocessor system. The global history of the directory comprises multiple sets of previous readers processed by the directory, with the total number of sets of previous readers corresponding to a designated history depth associated with generation of the set of predicted readers. The prediction process may use additional information in conjunction with the directory information, such as a designated subset of cache address information, processor node identification information, or program counter information.