Apple wins reversal in University of Wisconsin patent lawsuit
Apple Inc persuaded a federal appeals court on Friday to throw out a $234 million damages award in favor of the University of Wisconsin-Madison's patent licensing arm for infringing a patent on computer processing technology.
In a 3-0 decision, the U.S. Federal Circuit Court of Appeals said Apple deserved judgment as a matter of law, because jurors could not have found infringement based on evidence introduced in the liability phase of a 2015 trial.
The Wisconsin Alumni Research Foundation, or WARF, sued Apple in 2014, saying processors in Apple's iPhone 5s, 6 and 6 Plus smartphones infringed a 1998 patent describing a means to improve performance by predicting instructions given by users.
"We hold that no reasonable juror could have found literal infringement in this case," Chief Judge Sharon Prost wrote for the Washington, D.C.-based appeals court.
Also at The Verge and Engadget.
In a related development, the US International Trade Commission (ITC) has ruled that they will not be imposing an import ban:
A bit over a year ago, Qualcomm started the process of suing Apple at the US International Trade Commission (ITC) over alleged patent infringement. At the time, Apple was accused of violating six Qualcomm patents, ranging from power-saving technology to processor design. And while the case is far from over, according to Reuters the ITC has delivered its initial determination, finding that Apple has violated one of the six patents. Importantly, however, the ITC has also ruled that they will not be imposing an import ban, as Qualcomm originally requested.
[...] The ITC, in turn, has ruled today as part of its initial determination that Apple has indeed violated a Qualcomm patent, albeit just one of those remaining patents – what Reuters calls “related to power management technology”. We’re waiting on the ITC to publish the formal decision in order to confirm which specific patent it was, as all three remaining patents are related to power efficiency.
Previously: Apple Faces Hefty Damages After Losing Patent Lawsuit to Univ. of Wisconsin
Apple Ordered to Pay $506 Million to the University of Wisconsin in Patent Case
(Score: 4, Informative) by TheRaven on Sunday September 30 2018, @11:17AM (3 children)
This judgement is saying that the patent would cover keeping a hash table of loads with values associated with them and evicting the old value in cases of conflict, but would not cover Apple's version which (in common with most branch predictor state) aliases multiple loads. That's a somewhat interesting interpretation. As a person skilled in the art who has implemented a branch predictor for a research processor, I would disagree with this interpretation. It comes down to two possible designs, one that associates a prediction with multiple loads and one that associates the prediction with a single load but produces no prediction for a lot of loads. Reading the patent with the plain meaning of the word 'particular' covers only the second. I would be very surprised if this is what the authors of the patent intended, but it's what the court has decided that they wrote.
I'd thoroughly recommend reading the full opinion. I don't agree with the conclusion, but it clearly explains the thought process.
sudo mod me up
(Score: 1, Interesting) by Anonymous Coward on Sunday September 30 2018, @11:39AM (1 child)
Doesn't this just give more weight to the idea that software should not be patentable
How is this progress?
(Score: 2) by TheRaven on Thursday October 04 2018, @01:13PM
sudo mod me up
(Score: 2) by RamiK on Sunday September 30 2018, @01:36PM
I did and it's irrelevant:
That's what I meant by 99.9%. Consider how it only takes one SoC out of billions sold, each executing trillions of trillions (upon trillions...) of instruction over their life time to infringe the patent regardless of any particular interpretation (point intended). This is why I skipped all that convoluted legalese dictionary wank-off and suggested the judges can't count. Even with this most "liberal" interpretation of the patent's language, it's just statistically impossible Apple's predictor never branched based on a single load instruction. It doesn't even take much effort to write a parser-like switch statement that fills up the pipeline width/depth and run it over a piece of text input only to hit the token on the first case and satisfy the requirements. And since optimized compilers (and especially JITs) would break apart if...then blocks around the cache which is tuned around the pipeline, you're basically guaranteed to see this in real life even when browsing the web as those htmls and networking packet are being parsed.
compiling...