Slashdot Mirror
Graphics Rendering Patent Suits Target Apple, Samsung, HTC, RIM, LG and Sony
angry tapir writes "Formerly known as Silicon Graphics, Graphics Properties Holdings has filed six separate patent cases against Apple, Samsung, Research In Motion, HTC, Sony and LG with the U.S. District Court for the District of Delaware. The patent at issue in the lawsuits relates to floating point calculations to render graphics, and is registered as patent number 8,144,158 with the U.S. Patent and Trademark Office."
You've only read the damn abstract haven't you?
How many times do people have to say READ THE CLAIMS before it sinks in that the abstract normally only gives an example and the patents claims normally go far beyond that. In some patents the abstract comes close to being completely misleading. Incidentally the claims are not restricted to 16 bit floating point representations or any other size of floating point accuracy, plus it's a continuation of other patents, so don't forget to read their claims too.
Donte Alistair Anderson Roberts - hi son!
Karma: Chameleon
Sort of. 32 bits is not obvious. The IEEE standard for this stuff is actually pretty fucking complicated once you realize how much numerical analysis goes into the design. IEEE standardized on a set of formats, and half float is just a variant on those (e.g. the exponent bias is still 2^(E-1)-1). I templatized this for a compiler once - you could have a float with any number of mantissa bits and any number of exponent bits. Shit, should have patented it.
Plus, this patent is not a patent on half float, it's a patent on using floating-point AT ALL within a GPU. Talk about homesteading the noosphere :)
*Lowers the grammar nazi shotgun*
This seems to be the norm these days - a company that was formerly massively successful begins to die down and fade away, so in a last ditch attempt to cling to live they sue anyone and everyone that has ever done anything remotely similar to them.
Every successful company usually becomes that way under a founder. Or a founder-like figure (maybe the company was obscure for 100 years and then takes the world by storm - the leader in this case is like a founder). Under the founder all is well, and the company generally makes net-positive contributions to society.
Then the founder retires, and his hand-picked successor takes over. They usually start having more of an eye towards whatever the founder hired them for (often marketing, or finance, or whatever). However, they were mentored by the founder and usually are fairly true to the original dream.
After that the next succession is managed by the board's CEO search committee, and everybody after this could care less about visions and dreams, and instead aim to min/max their balance sheet and bonus check. Companies don't sue people - their leaders do. By the time a company reaches the state SGI is in, nobody who had anything to do with creating anything of worth is in charge.
Going by this logic, it does also mean that Apple must be ready to die soon. I mean, they can't have that much money, can they?
Every company is doomed to follow this cycle - it is the nature of wall street. Apple is now operating under the hand-picked successor. He will probably do reasonably well, but one day he will retire. Everything after that will be inertia. Oh, it takes a long time for a huge company to fail, and sometimes you get lucky and the wall street pick might actually turn out to be visionary. However, by-and-large the only innovations at Apple starting with the next CEO will be in the balance sheet.
Sooner or later you can only cut the bottom line so much before the fall in the top line starts. At that point the company will bleed off anything of worth it still has left, until its only function can be that decreed by law - it is still able to collect money, write checks to the decision-makers, and file lawsuits, shielding the decision-makers from personal liability. It is nothing more than a front at that point, but it will continue on.
Apple has always been lawsuit-happy, so who knows - perhaps we won't even last another CEO before the slide starts. It all depends on whether they start rewarding the lawyers more than the innovators like they do at most companies.
So patent trolls are terrrrrists? Interesting theory, let's see...
Hate us for our freedom to produce and innovate? Check.
Cripple the economy? Check.
Try to influence politics by their practices? Check.
I think you're on to something here...
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Here's an interesting article from Nov 2010 Embattled Silicon Graphics Portfolio Now On The Warpath
Yesterday, thanks to PriorSmart‘s Daily Litigation Alerts, I noticed Dell, HP and Lenovo all targeted in the same Delaware patent lawsuit by Graphics Properties Holdings, Inc. The titles of the two patents at issue were both “Display system having floating point rasterization and floating point framebuffering” (6,650,327 and 7,518,615). It sounded a bit familiar, so after a little searching, I found out that, sure enough, just a few days before the same company filed suit on the same 2 patents against Apple, Nintendo, Sony, Toshiba and Acer in the Southern District of New York. That said, I still wasn’t satisfied that I had correctly identified the source of my recollection. (You don’t often forget a term like “rasterization.”)
After a little more searching, I came across an older case, Silicon Graphics v. ATI Technologies, which had gone to trial in Madison, Wisconsin in 2008. The case was a close to a total loss for SGI, with Judge Barbara Crabb ruling that co-defendants ATI and AMD did not infringe the ’327 Patent, and that both defendants were authorized for certain uses under a license to Microsoft. So why are these new lawsuits being filed, and who is Graphics Properties Holdings? Graphics Properties is essentially what’s left of SGI after filing bankruptcy last year. (That’s right, again.) As for why these former SGI patents are now being asserted again, a court of appeals decision from earlier this year may help explain. Chief Judge Rader, in a unanimous opinion, undid just about everything that Judge Crabb had done.
Because the district court erroneously construed two of the three contested limitations in the ’327 patent this court vacates the summary judgment on claims with those terms. This court also determines that the district court erred with respect to the effect of the Microsoft license on direct infringement. * * *
[As a result,]this court vacates the district court’s non-infringement ruling and remands for consideration in light of the correct construction.
In other words, because Judge Crabb misinterpreted the meaning of critical terms in the patent, the ultimate conclusion of non-infringement was incorrect. Specifically, the phrase “a rasterization process which operates on a floating point format” was interpreted by Judge Crabb as requiring that the process “as a whole” needs to operate on a floating point format. It was undisputed that the accused products performed some rasterization processes on a floating point format, but others using fixed point values. Based on this construction, Judge Crabb (correctly) concluded that the accused products did not exactly match the claimed invention.
However, on appeal, Judge Rader noted that the specification recites a number of different rasterization processes, and that the patent claim uses the indefinite article a when describing rasterization on a floating point format. The correct construction, according to Judge Rader, is that “one or more of the rasterization processes (e.g., scan conversion, color, texture, fog, shading) operate on a floating point format.” Because it was also admitted that some of the rasterization process did use a floating point format, a judge simply can’t deny the patent holder its opportunity to prove infringement of the patent to the jury.
The contrast between these two constructions is dramatic, as potential design around opportunities for Judge Crabb’s narrower interpretation are significantly easier than for Judge Rader’s broader construction. Having emerged from this first battle with a broader interpretation of the patent claims, Graphics Properties has apparently decided to turn up the heat and pursue an even broader class of targets, including PC and game console manufacturers, and to do it on multiple fronts.
This is one of the stupidest patents I've ever heard of.
Even back in the fall of 1986 at University of Saskatchewan in my graphics class, the algorithms we started with were presented as floating point algorithms. We were then shown the integer variants on those algorithms, which we were told bluntly were used only because they were faster than floating point emulation.
So they got a patent for doing something that we were told not to do purely for performance reasons, not for any reason of logic or functionality. There is absolutely no doubt in my mind this whole patent should be overturned.
And, yes, I was heavily into computer graphics at the time. I even was a contributing publisher to a paper on the "Fast Line Clipping" algorithm, which really, in retrospect, was not so much an innovation as an example of a very advanced special case of loop and conditional unrolling that some of the more advanced modern compilers can probably to automatically at this present time. If you want to check out that crufty old article, you'll have a better chance of finding it by searching for Yang or Pospisil, the professor and grad student for the project; I was just a fourth year programmer at the time.
No, we didn't patent our algorithm. Back then the point of research and development was to learn and share, not to squat and sue.
I do not fail; I succeed at finding out what does not work.