Appeal 2007-1573
Application 10/705,094
portions of the image of the pipelines based at least in part on the analysis of
the time taken by each pipeline as claimed (Br 6).
Appellants emphasize that the Examiner’s assumption that a first
processor with a higher workload than a second processor will take longer to
complete such workload ignores a variety of other factors that affect the time
to complete a computing task. According to Appellants, these factors
include (1) memory latency, (2) different processor speeds in different
pipelines, (3) varying speeds in any one processor, (4) bus architecture, (5)
type of data structure, etc. Although Appellants concede that the Examiner’s
assumption may be true for a single processor, such an assumption is not
necessarily true when comparing two or more processors (Br. 6; Reply Br.
4).
The Examiner generally agrees that these other factors can affect
processing time. Nevertheless, the Examiner contends that the multiple
processors in Narayanaswami are not different types of processors (i.e.,
“heterogeneous”). The Examiner reaches this conclusion not from any
express disclosure in Narayanaswami, but rather from an assertion that it is
well known that designing homogeneous type multiprocessors is “much
easier” than heterogeneous type multiprocessors. The Examiner also asserts
that the speed of homogeneous multiprocessors increases proportionally
with the number of processors whereas the speed of plural heterogeneous
processors varies with the individual processors (Answer 5).
We will not sustain the Examiner’s rejection of representative claim
44. In short, we fail to see how Narayanaswami’s system analyzes the time
taken by each graphics pipeline, let alone automatically adjust the respective
portions of the image based on this time-based analysis as claimed.
4
Page: Previous 1 2 3 4 5 6 7 8 9 Next
Last modified: September 9, 2013