BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20111114T163000Z
DTEND:20111114T200000Z
LOCATION:
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: Linked data structures (LDS) such as linked lists, trees, graphs, and hash tables, are used heavily =0Ain programs. They utilize pointers and rely on pointer chasing to traverse the data structures. =0ATraditional loop-level parallelization is ineffective for LDS due to loop-carried dependence in the =0Atraversal loops. Consequently, a different parallelization strategy is needed. Parallelization can be =0Apursued at a higher level, among the LDS primitives, such as traversal, node insertion, and node =0Adeletion. Pursuing parallelization at the LDS primitive level, however, requires a different way to =0Areason about the correctness of operations, and requires fine-grain locking along with all its =0Aintricacies. LDS parallelization is a rarely covered topics in SC, and the aim of this tutorial is to =0Acover techniques that can be used to achieve an effective parallelization strategy for LDS.
SUMMARY:M02: Parallel Programming Techniques for Linked Data Structures
PRIORITY:3
END:VEVENT
END:VCALENDAR