Appeal No. 2001-2553
Application No. 08/512,369
each thread and the processor which is initially allocated to the
thread.
Representative independent claim 50 is reproduced as
follows:
50. A method for scheduling the execution of a plurality of
threads on a plurality of processors in a computer system,
wherein at least one of the threads can make more than one sleep
request, said method comprising the steps of associating a local
queue of threads with each of the processors; selecting movable
threads from the local queues; and on each of the processors,
performing the step of yielding control of the processor to a
thread that is selected from at least the selected movable
threads, wherein said step of selecting movable threads comprises
identifying a busiest processor among the plurality of
processors, the movable threads being selected only from eligible
threads in the local queue of the busiest processor, and wherein
said identifying step comprises identifying as a busiest
processor a processor which has received the smallest number of
sleep requests of any of the processors during a sampling period.
The examiner relies on the following references:
Belo 5,379,428 Jan. 3, 1995
Valencia GB 2,284,081 May 24, 1995
Rudolph et al., “A Simple Load Balancing Scheme for Task
Allocation in Parallel Machines”, Dept. Of Computer Science,
Hebrew Univ. Jerusalem, Israel, 1991, pp. 237-245. (Rudolph)
Leung, “An Execution/Sleep Scheduling Policy for Serving an
Additional Job in Priority Queuing Systems”, Journal of the
Association for Computing Machinery, Vol. 40, No. 2, April 1993,
pp. 394-417.
-2–
Page: Previous 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Next
Last modified: November 3, 2007