Docket No. 2017-2265, -2380
PROST, BRYSON, O’MALLEY
September 28, 2018
Brief summary: DC holding that Apple infringed WARF’s patent reversed. DC grant of SJ of no anticipation affirmed.
Summary: Apple appealed DC holding that it infringed WARF’s US 5,781,752 (expired on Dec. 26, 2016) relating “to how computer processors execute a computer program’s instructions” using “prediction” (the processor makes an informed decision whether to speculate if “out-of-order execution of instructions” to increase performance) and award of over $234 million in damages. Apple also appealed the DC grant of SJ to WARF of no anticipation of the asserted claims. WARF accused Apple’s A7, A8, and A8X integrated circuit chips, which include one or more processors” that “include a Load-Store Dependency Predictor (‘LSD Predictor’).” Apple argued that its processors do not literally infringe “the ‘particular’ nor the ‘mis-speculation’ limitations” of the asserted claims. The “particular” limitation of independent claim 1 “requires a predictor that ‘produce[s] a prediction associated with the particular [load] instruction”, and independent claim 9 requires a prediction table that “create[s] an entry listing a particular [load] instruction and [store] instruction each associated with a prediction.” Apple and WARF disputed the meaning of “particular” with WARF arguing Apple’s expert “should have been forbidden from making any suggestion that each prediction must be associated with one and only one load instruction” and Apple arguing for “its plain and ordinary meaning” which was consistent with its expert’s testimony. The DC agreed with Apple and found that claim 1 “contemplates a single load instruction…consistent with the plain meaning of the claim terms ‘the’ and ‘the particular’” (Phillips, FC 2005 (“the ‘ordinary meaning’ of a claim term is its meaning to the ordinary artisan after reading the entire patent”)). The FC panel found this reading did not “read out the preferred embodiment” (FN7) and found. The FC panel also concluded, “drawing all reasonable inferences from the evidence in favor of WARF”, that “no reasonable jury could have found literal infringement in this case” because using “Apples hashing algorithm…multiple load instructions may hash to the same load tag” and “therefore be associated with a group of load instructions” (“the dispositive issue is whether other instructions also hash to that tag, such that the prediction is associated with a group of instructions, rather than a particular instruction”; “insufficient evidence to support WARF’s theory that Apple’s load tags are sometimes associated with a single load instruction” (Broadcom, FC 2013 (“a product that ‘sometimes, but not always, embodies a claim nonetheless infringes”)). In FN8, the FC panel explained its disagreement with WARF’s argument that the use of “comprising” in the preamble “allows for additional, unrecited elements” since “‘[c]omprising’ is not a weasel word with which to abrogate claim limitations” (Spectrum, FC 1998). The FC panel therefore found insufficient evidence of literal infringement and set aside the jury’s finding. In its grant of SJ to WARF of no anticipation, the DC found “the Steely prior art reference…did not disclose the ‘prediction’ claimed”. The FC panel affirmed the DC’s grant of SJ under the “ordinary meaning” of “prediction” (GPNE, FC 2016 (“[w]here, as here, ‘a patent ‘repeatedly and consistently’ characterizes a claim term”, it will be construed in that way)).