Ex parte Brunner
Appeal 2009006904; Appl. No. 10/877,358; Tech. Center 2600
Decided February 14, 2011
A representative claim on appeal read:
1. A method to generate a display-wide visual effect, comprising:
copying an image buffer’s contents into a second buffer;
filtering the second buffer’s contents back into the image buffer using a graphics processing unit to generate a specified visual effect, wherein the image buffer is associated with a system frame buffer; and
compositing an application-specific window buffer into the image buffer, wherein the act of compositing is performed by the graphics processing unit after the act of filtering.
During prosecution, the Examiner took the position that "any processing unit that carries out graphical operations, such as outputting display data to a display device, is a graphic processing unit." The Examiner used this interpretation in an obviousness rejection, asserting that "because the at least one programmable processor in Hamburg is described as outputting data to at least one output device (such as the described LCD screen), the at least one programmable processor described in Hamburg is considered a graphics processing unit."
The Applicant contested this interpretation, and submitted an IEEE article titled "GPU Computing" as evidence of the meaning of "graphic processing unit" to a POSITA. The Applicant referred to this article in the Appeal Brief, arguing:
As one of ordinary skill in the art would recognize, the use of a graphics processing unit (GPU) entails a significantly different process than using a system's central processing unit and providing for efficient use of display resources. "[T]he architecture and programming model of the GPU are maredly different than most other commodity single-chip processors." (Owens et al., GPU COMPUTING ...; Evidence Appendix.) "The input to the GPU is ... triangles ... [each] triangle generates a primitive called a "fragment" at each screen-space pixel location that it covers. Because many triangles may overlap at any pixel locations, each pixel's color value may be computed [by the GPU] from several fragments." (Id.) In contrast, a central processing unit is driven by standard programming commands and process a single pixel at a time. Thus, the use of a dedicated graphics processing unit as claimed is not an insignificant difference between the cited art and the claimed invention.
The Examiner elaborated on his position in the Answer:
[T]he Applicant's claims don't limit the disclosed graphics processing unit to the graphics processing unit described by Owens et al. The claimed graphics processing unit is only limited to a graphics processing unit that [functional language of the claim.]
The Examiner also pointed to a portion of Hamburg that taught:
The invention can be implemented advantageously in one or more computer programs that are executable on a programmable system including at least one programmable processor ... Suitable processors include, by way of example, both general and special purpose microprocessors.
The BPAI found fault with the Applicant's interpretation of "graphics processing unit":
Appellants have not demonstrated error in the Examiner’s conclusion that the recited “graphics processing unit” is broad enough to read on Hamburg’s “programmable processor.” The reason is that Owens describes examples of GPUs without offering what can be accurately characterized as a broadest reasonable definition of that term.6 Furthermore, Appellants’ reliance on Owens’ discussion of “triangles” and the “fragments” generated thereby is misplaced in view of Owens’ description of triangles as “typical” rather than as essential: “The input to the input to the GPU is a list of
geometric primitives, typically triangles, . .” (emphasis added). Owens at 880, 2d col., 1st para. under heading “A. The Graphics Pipeline.”
The Board found that the Applicant did not demonstrate error in the Examiner's interpretation, then used the Examiner's interpretation to conclude that Hamburg did disclose a graphics processing unit. The Board affirmed the obviousness rejection.
My two cents: The Applicant was smart to zero in on the real issue here: the meaning of GPU. And to hit the issue head on by introducing evidence.
Unfortunately for this Applicant, looks like this Board wanted a true definition, where the article was instead a discussion of GPUs. I think the Board made a mistake by not looking at the article as a whole and what it said about how POSITAs interpret "graphics processing unit." To me, the article as a whole clearly taught that GPU means something other than a processor that carried out graphics operations, by explaining how GPUs are different than regular microprocessors. The Applicant could have done a better job of explaining this – and probably would have had he known the Board was going to focus on this.
I haven't read many BPAI opinions dealing with the use of technical articles rather than dictionary definitions to argue meaning of a claim term to a POSITA. So I'm not sure how common this outcome is. My sense is that a true definition from a dictionary is better.
I'm disappointed the Board didn't recognize that the Examiner's interpretation was ridiculously broad. Yeah, okay, maybe the Applicant didn't have the best argument, and it's true that a GPU isn't limited to triangle primitives. But the meaning of GPU isn't as broad as a processor that carries out graphics operations. My PC in 1990 "carried out graphics operations" in order to put graphics on the screen, and it didn't have a GPU.
Maybe the Applicant was dead in the water here anyway, thanks to the offhand statement in the reference that "[s]uitable processors include, by way of example, both general and special purpose microprocessors." A graphics processing unit is certainly a special purpose microprocessor, so this blanket statement in the reference may amount to a teaching of the claimed element.
Looks like the Applicant got away with a procedural error: as far as I could tell, the GPU Computing article was submitted for the first time in the Evidence Appendix of the Appeal Brief.
Finally, I note that decision is not as big a travesty as the Board's opinion in Ex parte Giacalone, There, the Board found that because "digital signal processing" meant manipulating images or other real-world signals, a video processor was a DSP because it processed images. (I posted long ago about Giacalone; read about it here.)