Appeal 2007-2127
Reexamination Control No. 90/006,621
In preemptive multithreading, Thread 1 is interrupted at the end of its
timeslice, the Thread 1's context is saved (including Thread 1's program
counter), Thread 2's context is retrieved and loaded into the CPU (including
Thread 2's program counter indicating the next instruction to be executed),
Thread 2 executes until the end of its timeslice, whereupon Thread 2's
context is saved, Thread 1's context is retrieved and loaded into the CPU,
and the process repeats. Each thread executes an incremental portion of its
task during each timeslice. At any point in time, both threads are executing
at some point between their beginning and end, except for needing to share
the CPU—this is what is meant by concurrent execution of threads.
In the 1982 application, the compiler is interrupted, registers are
saved, the editor interrupt service routine executes until it finishes, the
registers are restored, and control is returned to the compiler by an
instruction in the interrupt service routine. According to what Patent Owner
states to be the normal operation, the editor finishes before the end of the
timeslice and control returns to the compiler. As seen in the middle figure,
the compiler is then interrupted at the next interrupt. Because of the varying
amounts of time required by the editor, depending on the operation to be
performed and whatever else is happening on the computer, the editor and
compiler do not execute for equal periods of time. If the editor takes longer
than a timeslice, the editor continues to execute past the interrupt time since
interrupts are disabled as shown by the bottom figure. The editor is never
interrupted but always finishes executing and returns control to the compiler.
Importantly, the compiler and editor do not execute concurrently as
that term is defined in "multithreading." Concurrent execution requires that
72
Page: Previous 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 Next
Last modified: September 9, 2013