Nvidia Sues Qualcomm, Samsung Over GPU Patents

[Exophase]

What is your point though? A patent is a patent is a patent. Wouldn't you agree? If a patent is truly found to be infringed upon, then restitution should be agreed upon and paid. If it is found that no patent was infringed upon, then a new suit should be brought about to pay for the legal fees incurred from initial suit. IMHO.

There are several criteria a patent needs to be valid. Successfully being filed doesn't mean that the patent can't be found invalid in court. The patent office simply doesn't have the means to properly and thoroughly audit every patent of this nature.

Look at the criteria here:

http://www.bitlaw.com/patent/requirements.html

The publication one is big minefield, there are a bazillion publications made on graphics rendering techniques, hardware or software.

well going this route, nothing can receive a patent as new tech is always an evolution of prior art...

It has to both be novel and offer a non obvious improvement over prior art. It's a really tough sell to call die integration a non obvious improvement when Moore's Law has been driving that for everything imaginable for a long time.

But between you an me, when Geforce256 aka NV10 came to the market, with hardware T&L, it was revolutionary. Nothing was like that at that time. Same for NV30 with hardware programmable shaders (even if this GPU suffered from many flaws). Each time, competitors came very to the market with their similar tech.
You may disagree with nvidia claims but nobody will argue against the prominent role they played in the graphic technology evolution. They invest heavily in R&D, they patent their ideas first and they put it in real products...

You think it's revolutionary because you only care about the impact it had on the PC gaming market. When it came out you probably didn't have the first clue that the technology was already being used in workstations and consoles, but that doesn't change anything. Nothing in any of the patents make a claim about novelty by way of being applied to a particular market (which would probably be a weak claim, but I digress)

With products like GPUs and CPUs the devil is in the details. It's not the high level ideas that mean anything (which have very often been hashed out in academia well ahead of time), but the implementation. How they balance feature set vs die and area budget, how they tune and size and optimize everything, how they write the drivers - it's things like this where the value of their work lies. And I won't deny for a minute that nVidia has tremendous strength in the quality of their designs (although they hardly have it alone). But these patents are about much broader and higher level ideas that they shouldn't have claim to, and I don't see an argument that they contributed to the strength of their competitors GPUs at all.

 

[tviceman]

Nvidia said they started talks with Qualcomm in 2012, but the first Adreno GPU was released in 2008, so why did Nvidia wait 4 years before they decided that they were infringing?

Probably because of the Intel settlement. Intel and Nvidia probably knew that had Intel not settled out of court over locking Nvidia out of the chipset business, that Intel would have had to pay royalties for GPU patents. Given that information, Nvidia decided it would be wise to pursue this with other companies.

Given that Microsoft gets a royalty for every Android phone sold, that Qualcomm gets a royalty for every cellular modem sold, that Apple gets a royalty for a rectangle with rounded corners, and that Rambus gets a royalty for anyone sneezing, it seems reasonable that Nvidia would pursue this avenue as a revenue stream as well.

Nvidia is already paying royalties to Qualcomm for their iceria modems. Tit-for-tat.

 

[Exophase]

Nvidia said they started talks with Qualcomm in 2012, but the first Adreno GPU was released in 2008, so why did Nvidia wait 4 years before they decided that they were infringing?

Because that's what these companies do, over and over again. They file patents for everything under the sun using the broadest language possible and clearly drawing upon established ideas. Then they knowingly wait while their competitors use technology they feel is covered by the patents, until that competitor amasses a huge ecosystem and revenue stream from the technology and is utterly dependent on it. Then they sue for a big chunk of that success. If they did the honest thing and filed against the company as soon as they were aware of an infringement they'd make much less money. People are accusing Qualcomm for taking advantage of the work nVidia put into their ideas (which I find highly questionable), but it seems to me that nVidia is taking advantage of the work Qualcomm put into making their products successful. Which, btw, has to do with a lot more than just a few GPU technologies that they use.

 

[sontin]

Seriously?
Qualcomm is advertising their GPU as a feature. If you think they would be as successfull as without it maybe you should tell them this.

Using this logic why should anybody pay them license fees for their modem technology?

 

[AnandThenMan]

I hope this burns Nvidia big time, so sick of the patent trolling.

 

[Keysplayr]

Exophase, can you please fix post 51 as I didn't say the things in your 2nd and 3rd quotes. Thanks.

And as for the first quote, so we are calling all these Nvidia patents invalid for the time being? What would you have us believe and why?

 

[Mopetar]

I think that there's a lot of misunderstanding in regards to how patents work.

There may well be examples of technology related to the things that NV has patents on that existed well before NV's patents, but that doesn't necessarily make them invalid. Let's pick on unified shaders.

If at some point NV found a new method of creating a unified shader that resulted in a massive performance gain, they could patent that method. Just like if Ford were to find a better way to making a more efficient engine, they could patent that even though engines have been around for a long time. If every other company suddenly started using the same method as NV only after it was developed and patented by NV, then NV would have a pretty good case.

I haven't read the patents themselves yet, so I can't say whether NV is trying to claim a particular implementation of an idea or the idea itself, but simply pointing to an earlier implementation of an idea isn't a good defense against a patent that covers a superior implementation of that idea.

 

[Exophase]

Seriously?
Qualcomm is advertising their GPU as a feature. If you think they would be as successfull as without it maybe you should tell them this.

I didn't say anything like that, I said that a large part of their success is due to factors outside the GPU. Not that NONE of it is.

Using this logic why should anybody pay them license fees for their modem technology?

You seem to be under the impression that I think all patents and claims for licensing are invalid, when that's not the case. I think these patents are too broad. Quite a bit different from something as specific as defining a communications protocol.

Exophase, can you please fix post 51 as I didn't say the things in your 2nd and 3rd quotes. Thanks.

Sorry about that, I fixed it.

And as for the first quote, so we are calling all these Nvidia patents invalid for the time being? What would you have us believe and why?

I wouldn't make such a strong statement, and I certainly wouldn't say that the courts will see it this way. Simply that from what I've seen of the patents I have reason to be skeptical that the other parties are in violation of anything specific enough that wasn't well established in older products and public research.

I think that there's a lot of misunderstanding in regards to how patents work.

I don't think so, unless you argue that the page I linked earlier is wrong in its criteria for patents. You're right, you can come up with a novel new implementation of something and defend that. But I don't think the infringements nVidia is claiming have much to do with that.

 

[rgallant]

per the filing

they have these chips made off shore then import them to the US

they have these chips made off shore where we have ours made and then we import them also to the US --fixed

 

[Cookie Monster]

I don't think so, unless you argue that the page I linked earlier is wrong in its criteria for patents. You're right, you can come up with a novel new implementation of something and defend that. But I don't think the infringements nVidia is claiming have much to do with that.

The thing is that you don't just go and file a lawsuit for patent infringements to the likes of the Samsung and Qualcomm if what your suggesting is correct. I think your being very ignorant imho because a) back 15 years ago, those simple concepts/inventions/fundamentals weren't so simple so its quite amusing to imply that even your pet dog knew about these is laughable and b) don't be fooled by the patent titles.. its all about the claims in those patents.

Not all unified shaders are the same, not all rendering pipeline are the same. Theres something (often very very low level) that could effectively make it a legitimate invention which gives you alot of benefit. Its sometimes mind blowing to know just how many clever things ala inventions are implemented in modern day technology.

Like I've said earlier, they will most likely have done their homework e.g. prior arts, when the patents were filed, what claims are infringed upon, potential cross licensing deals that were infringed upon (e.g. when Qualcomm acquired AMD's handset division most likely any patent agreements here would've been terminated) etc.

This sets out to be an interesting case because it isnt another rambus case and its not about rectangles with round corners.

And yes I think the patent system is a joke especially how most patents are "land grabs" and too general but I dont think this applies to this case.

 

[dguy6789]

Tegra can't compete with Qualcomm so they try and make some money citing illegitimate patents. Nvidia really is an integrity free company.

 

[Mopetar]

But I don't think the infringements nVidia is claiming have much to do with that.

I haven't seen any of the news stories actually list the patents, so without reading them I can't comment further. Obviously Qualcomm, et al. are going to claim that they don't infringe or that the patents aren't any good, but without having read them yourself what confidence can you really have in your statement?

What anyone here might think about the case, especially with potentially uninformed opinions, isn't going to amount to any use. If someone wants to go through the patents and point out some prior art, by all means do so, but simply suggesting that it exists or that something like it exists isn't very helpful.

 

[xpea]

Mopetar, the infringed patents are listed in the register article:
http://www.theregister.co.uk/2014/09/04/nvidia_sues_qualcomm_and_samsung/

 

[Exophase]

Not all unified shaders are the same, not all rendering pipeline are the same. Theres something (often very very low level) that could effectively make it a legitimate invention which gives you alot of benefit. Its sometimes mind blowing to know just how many clever things ala inventions are implemented in modern day technology.

Yeah, could is the operative word here. You're acting like I completely dismissed the possibility as soon as I read the charges, when I actually spend some time looking at the patents. Now, patents can be very confusing and I'm sure patent lawyers can dig all sorts of technicalities out of them given enough time, so I'm not saying nVidia definitely has no case here, but some of you on the other hand are convinced that nVidia must have a case because otherwise they wouldn't be claiming anything, right? As if patent infringement claims have never been dismissed in court?

There are low level examples of nVidia's particular implementation, yes, but they don't match up with low level implementations of their competitors. This is evident if you look at the mobile GPU architectures in any depth (including whatever reverse engineering has been made on them). Several of the patents don't even really correlate with modern hardware at all, for instance referencing translation and lighting modules, where it would be a massive stretch to say that GPUs today still have such a thing. nVidia is absolutely staking their claims on some very broad high level concepts, not implementation details.

And all you have to do to show that "everyone and their dog" knew that this is the direction graphics was heading in is to look at publications from the 80s and 90s.

This sets out to be an interesting case because it isnt another rambus case and its not about rectangles with round corners.

And yes I think the patent system is a joke especially how most patents are "land grabs" and too general but I dont think this applies to this case.

Funny how nVidia is given the benefit of the doubt without question while Rambus is dismissed so readily, I wonder why that is, and I wonder what actual reasons you have for being so confident about these patents not being too general...

 

[ams23]

And? SGI is not defending their patents here, nor is nVidia defending patents they acquired from SGI.

You missed the point. History has shown that NVIDIA, as a major industry force in pushing forward graphics and visual computing research and development, is more than willing to enter into graphics patent cross-licensing agreements with the major players in the industry. They have done this with SGI. They have done this with AMD. They have done this with Intel. All the major players today in the desktop/notebook space who are looking to push forward modern day graphics have a graphics patent licensing or cross-licensing agreement with NVIDIA. And yet, due to sheer arrogance, Samsung is claiming essentially zero responsibility for the graphics IP that they productize and bring to market, while Qualcomm is simply ignoring the issue over graphics IP altogether.

It has to both be novel and offer a non obvious improvement over prior art.

To say a "novel" and "non obvious" improvement is vague. Clearly these seven graphics patents (and the other 7000+ graphics patents in NVIDIA's portfolio) were granted for a reason. You can say what you want about the validity of patents and the patent process, but like most people here, you are not a patent lawyer nor are you a graphics hardware engineer who worked on said patents. Note that even Samsung does not appear to be challenging NVIDIA's patent infringement claims. Instead, they are passing the blame to their third party IP suppliers.

 

[monstercameron]

You missed the point. History has shown that NVIDIA, as a major industry force in pushing forward graphics and visual computing, is more than willing to enter into graphics patent cross-licensing agreements with the major players in the industry. They have done this with SGI. They have done this with AMD. They have done this with Intel. All the major players in the desktop/notebook spacce who are looking to push forward modern day graphics have a patent cross-licensing agreement with NVIDIA. And yet, due to sheer arrogance, Samsung is claiming essentially zero responsibility for the IP that they productize and bring to market, while Qualcomm is simply ignoring the issue over graphics IP.



To say a "novel" and "non obvious" improvement is vague. Clearly these seven graphics patents (and the other 7000+ graphics patents in NVIDIA's portfolio) were granted for a reason. You can say what you want about the validity of patents and the patent process, but like most people here, you are not a patent lawyer nor are you a graphics hardware engineer who worked on said patents. Note that even Samsung does not appear to be challenging NVIDIA's patent infringement claims. Instead, they are passing the blame to their third party IP suppliers.

wow you make it sound like nvidia is entitled to these cross licensing agreements. We will know soon enough if the claims that nvidia is a morally bankrupt company or rightly entitled soon enough.

 

[ams23]

wow you make it sound like nvidia is entitled to these cross licensing agreements.

NVIDIA has more fundamental graphics patents than any other company in the world (by a large margin too). Like it or not, the graphics patents cross-licensing agreements with AMD and Intel exists, so it is what it is. The difference in the ultra mobile space is that graphics IP suppliers and graphics IP integrators have essentially ignored the issue of graphics IP altogether as they pursue increasingly more modern GPU architectures.

 

[Abwx]

In a word, yes.

In a word, no, neither Intel, Motorola or Zilog could patent the use of integrated ALUs , registers and all pieces of a processor that were on different integrated circuits on previous designs, seemingly the addition inside the uprocessors of the once separated FP co processors wasnt patentable.

NVIDIA has more fundamental graphics patents than any other company in the world (by a large margin too).

That s obviously not true, the fundamentals patents are detained by the first firms that were on the market and that s ATI mainly, the bulk of Nvidia patents are the ones they got from 3DFx, for the rest they are mainly buying little firms as a mean to enhance their portfolio but are not a truly creative firm, check their page at Wiki.

 

[Cookie Monster]

Yeah, could is the operative word here. You're acting like I completely dismissed the possibility as soon as I read the charges, when I actually spend some time looking at the patents. Now, patents can be very confusing and I'm sure patent lawyers can dig all sorts of technicalities out of them given enough time, so I'm not saying nVidia definitely has no case here, but some of you on the other hand are convinced that nVidia must have a case because otherwise they wouldn't be claiming anything, right? As if patent infringement claims have never been dismissed in court?

There are low level examples of nVidia's particular implementation, yes, but they don't match up with low level implementations of their competitors. This is evident if you look at the mobile GPU architectures in any depth (including whatever reverse engineering has been made on them). Several of the patents don't even really correlate with modern hardware at all, for instance referencing translation and lighting modules, where it would be a massive stretch to say that GPUs today still have such a thing. nVidia is absolutely staking their claims on some very broad high level concepts, not implementation details.

And all you have to do to show that "everyone and their dog" knew that this is the direction graphics was heading in is to look at publications from the 80s and 90s.

I can't comment further because I don't know the exact details with regards to the patents that are supposedly infringed upon.. so we shall see?

Funny how nVidia is given the benefit of the doubt without question while Rambus is dismissed so readily, I wonder why that is, and I wonder what actual reasons you have for being so confident about these patents not being too general...

What has rambus done in the last decade other than filing patent infringement lawsuits to other companies? I can't think of any meaningful value they have added to the consumer/world in that time period. nVIDIA (and same goes for AMD/ATi) on the other hand has made what PC gaming or gaming in general possible today. So many things have changed in this area and its allowed gamers around the world to really play games that were once basic 2D to a full life like 3D experience. On top of that, mathematical problems that would take weeks, months to calculate would be done in mere days or hours thanks to GPUs which are now becoming more general purpose. This is huge for the world of academia and industrial companies. These are just a few from a long list of benefits or value that they have constantly added to its end users and pretty much to the world.

Yes, I think its fair to give that tiny benefit of the doubt given that they rarely go after other competitors via lawsuits if at all. And i can say the same back to you. What actual reasons have you for being so confident about these patents being too genera...l?

 

[Cookie Monster]

In a word, no, neither Intel, Motorola or Zilog could patent the use of integrated ALUs , registers and all pieces of a processor that were on different integrated circuits on previous designs, seemingly the addition inside the uprocessors of the once separated FP co processors wasnt patentable.

Probably because they were patented very long time ago before any of those companies existed where now the patents for these have all but expired and have become free knowledge. Almost all basic fundamental patents from my knowledge have been expired.

That s obviously not true, the fundamentals patents are detained by the first firms that were on the market and that s ATI mainly, the bulk of Nvidia patents are the ones they got from 3DFx, for the rest they are mainly buying little firms as a mean to enhance their portfolio but are not a truly creative firm, check their page at Wiki.

Actually it all depends on when they filed the patents and if any prior arts existed back then. Just because your the first company on the block doesn't necessarily grant this either. Bulk of nVIDIA patents are from buying/acquiring other firms? same goes for ATi.. qualcomm, samsung.. almost every company out there that has the resources to do so. And why not? I mean for example not many know that it was really ArtX's technology/patents that paved way for the glorious R300 NOT ATi. So lets not fool ourselves..

 

[Keysplayr]

Yeah, could is the operative word here. You're acting like I completely dismissed the possibility as soon as I read the charges, when I actually spend some time looking at the patents. Now, patents can be very confusing and I'm sure patent lawyers can dig all sorts of technicalities out of them given enough time, so I'm not saying nVidia definitely has no case here, but some of you on the other hand are convinced that nVidia must have a case because otherwise they wouldn't be claiming anything, right? As if patent infringement claims have never been dismissed in court?

There are low level examples of nVidia's particular implementation, yes, but they don't match up with low level implementations of their competitors. This is evident if you look at the mobile GPU architectures in any depth (including whatever reverse engineering has been made on them). Several of the patents don't even really correlate with modern hardware at all, for instance referencing translation and lighting modules, where it would be a massive stretch to say that GPUs today still have such a thing. nVidia is absolutely staking their claims on some very broad high level concepts, not implementation details.

And all you have to do to show that "everyone and their dog" knew that this is the direction graphics was heading in is to look at publications from the 80s and 90s.



Funny how nVidia is given the benefit of the doubt without question while Rambus is dismissed so readily, I wonder why that is, and I wonder what actual reasons you have for being so confident about these patents not being too general...

Exophase, you're being unreasonable. Nobody gives Rambus any benefit of any doubt because they have a LOOOOOOOONG history of bait and switch lawsuits. Rambus is synonymous with litigation. THAT is why Rambus is dismissed so readily. No other reason but a proper one. Nvidia isn't being given the benefit of the doubt because with the exception of you, and a couple of others here, there isn't anything to be doubted YET. Or confirmed, YET.

And I highly doubt that you know what you're looking for in those patents either. It's not like you could know exactly what you're looking for.
You say you carefully looked them over. What were you looking for? Be incredibly specific too. You must be incredibly specific and detailed in your response because Nvidia will and so will Qualcomm and Samsung.

And you must offer some sort of proof or reasoning why it isn't patent infringement and give reasons, exact detailed reasons, why the patent doesn't conflict with earlier already existing tech, as you put it.

So you see, when you won't, or more to the point, can't do this, we are back to square one and we are wondering if the law suits has any merit and wont know until the court proceedings are underway, or maybe even over.

 

[SPBHM]

something doesn't sound quite right for me because of who they are targeting and when...

also comparing this to the 1.5bi from Intel is crazy, since it involved a lot more.

 

[sontin]

That s obviously not true, the fundamentals patents are detained by the first firms that were on the market and that s ATI mainly, the bulk of Nvidia patents are the ones they got from 3DFx, for the rest they are mainly buying little firms as a mean to enhance their portfolio but are not a truly creative firm, check their page at Wiki.

What? 6 of the 7 patents where created by nVidia. Only the GigaPixel patent didn't come from them.

And btw: From which other company could they have bought graphics IP? :hmm:

 

[sontin]

I didn't say anything like that, I said that a large part of their success is due to factors outside the GPU. Not that NONE of it is.

They are successful because they can offer a whole package including an advanced GPU. And the GPU will be more and more important in the future.
Using the same logic nVidia shouldn't pay them any fees for their radio IP. Their Tegra line doesnt sell because of the LTE softmodem from Icera.

You seem to be under the impression that I think all patents and claims for licensing are invalid, when that's not the case. I think these patents are too broad. Quite a bit different from something as specific as defining a communications protocol.

I dont see how this patent ("GPU") is too broad: https://www.google.com/patents/US6198488

 

[Keysplayr]

I've also heard a couple of people say in this thread, that Nvidia cannot compete with Qualcomm and this is why the lawsuit is taking place. Is this true? What is, from Qualcomm, the latest most powerful SoC for smartphones and/or tablets released to date? Released mind you. Not rumored or expected. And also throw in what Apple has released currently. Not due for release but out now, not rumored or expected. All companies are obviously currently working on their next best thing.

Silverforce11 in post #6 "STFU NVIDIA, a disgraceful move. Can't compete against Samsung & Qualcomm with Tegra? Sue them."

Care to elaborate? I thought K1 was right up there with the best of them this gen?

DGuy6789 in post #61 "Tegra can't compete with Qualcomm so they try and make some money citing illegitimate patents. Nvidia really is an integrity free company."

Again, I thought K1 was competitive with Qualcomm's and Apples latest? Am I mistaken? How did you come to your conclusion that Nvidia cannot compete with Tegra? Did you mean another older version of Tegra that didn't do so hot? Or the latest K1?