Back to Article

  • shabby - Thursday, September 04, 2014 - link

    Mind = blown. Reply
  • Impulses - Friday, September 05, 2014 - link

    Oh boy, might as well have tossed Apple into the lawsuit as well so they could drag it out to 2020. Reply
  • garadante - Thursday, September 04, 2014 - link

    It'll be interesting to see what information comes to light and how all sides argue their case. I don't know enough to say whether this is a patent troll case like the Apple/Samsung case or if there is legitimate infringement going on. Reply
  • just4U - Thursday, September 04, 2014 - link

    I would say (for the most part...) that most companies out there simply license stuff from the big 3 (Nvidia, Amd, Intel) as they don't have much of a leg to stand on. These companies have so much in R&D out there that your bound to run into infringement on their IP.

    We don't hear much about it because it's always on the down low.. until now anyway. I can't see Intel/Nvidia/AMD being patent troll types.. specially the later two.. if one or the other is going after a company they likely have a very very good case.
  • tuxRoller - Thursday, September 04, 2014 - link

    img tech has been around for nearly as long as nvidia, and been focused on the mobile arena. This is a dangerous game for nvidia. Reply
  • ArthurG - Thursday, September 04, 2014 - link

    And yet Nvidia receives hundred of millions each year from Intel for the licence of the tech used by Imagination in their igpu. Nvidia won the case against Intel for same patents, they are no reason they won't win again... Reply
  • Khato - Friday, September 05, 2014 - link

    Actually, the NVIDIA case was entirely with respect to Intel somewhat violating their 2004 agreement with respect to chipsets when Intel began integrating such into the processor and effectively pushing NVIDIA out of that market. It had nothing whatsoever to do with the patents. (Just do a search for the 2004 Intel NVIDIA cross licensing agreement if you want the background, in fact the Anandtech article on the 2011 settlement even does a good job explaining such.) Reply
  • ArthurG - Friday, September 05, 2014 - link

    This is not true. From the Nvidia complain:
    "Intel is licensed by NVIDIA under the Asserted Patents and also conducts significant domestic industry activities in the United States relating to products practicing the Asserted Patents. As part of a cross-licensing agreement, Intel agreed to pay NVIDIA a total of $1.5 billion over five years for the use of NVIDIA’s patents and its state-of-the-art GPUtechnology"
  • Khato - Friday, September 05, 2014 - link

    You're ignoring what I said - the 2004 agreement between Intel and NVIDIA already covered cross-licensing of GPU technology. The disagreement between Intel and NVIDIA was with respect to that agreement - NVIDIA claimed that it allowed them to continue building chipsets for Intel systems regardless of interface while Intel claimed that it only allowed them to build chipsets using the AGTL+ bus that they were no longer using. Which was a rather weak argument by Intel, hence why they settled and paid NVIDIA that nice sum for 'breaching' the previous cross-licensing agreement while inking a new one.

    Sorry, but there's no way to spin the 2011 agreement as being anything more than a continuation of the graphics and CPU cross-licensing agreement already in place. Because, oh yeah, how many Intel patents would NVIDIA be stepping on with their Denver core? It's not a one-sided agreement like many wish to believe. And I suspect that's the same kind of agreement NVIDIA is angling for with Qualcomm - not money, just free reign in their designs without having to pay any royalties on the modem side.
  • djgandy - Friday, September 05, 2014 - link

    Imgtec (Videologic) has been around way longer than Nvidia. The on chip tiling one is interesting, surely covered by prior art. Reply
  • eSyr - Friday, September 05, 2014 - link

    Imagination, in fact, is 8 years older and moved to mobile after its tile-based rendering, albeit being more advanced in terms of minimizing overhead on complex scenes, have been beaten by brute force solutions from nvidia/ati (geforce 2 era, i suppose). Reply
  • tuxRoller - Friday, September 05, 2014 - link

    Sorry. You're correct. Wikipedia says 1985. Reply
  • Sttm - Thursday, September 04, 2014 - link

    I can't think of a company that is actually making products and investing heavily in R&D as a Patent Troll. I think that terminology should be saved for the many companies that do not actually do anything but hold patents and sue people.

    Nvidia is actively competing against these companies, and if their GPU designs infringe on Nvidia's patents, which Nvidia obtained by doing the R&D, then Nvidia should collect a licensing fee, or win a Jury pay out.
  • Samus - Friday, September 05, 2014 - link

    "I can't think of a company other than Apple that is actually making products and investing heavily in R&D as a Patent Troll."

  • dylan522p - Saturday, September 06, 2014 - link

    Apple's R&D is fucking massive and they have many legitimate patents. WTF are you talking about. Reply
  • Gigaplex - Sunday, September 07, 2014 - link

    They also have a bunch of weak patents and somehow get away with suing over them. I still find it hard to believe that patents essential to building the phones aren't allowed to be considered when only the look and feel patents count. Kind of defeats the point on making patents available as FRAND. Reply
  • name99 - Friday, September 05, 2014 - link

    "I don't know enough to say whether..."
    But you know enough to confidently assert that Apple/Samsung was a case of patent trolling.
    Got it --- you're clearly a dispassionate observer...
  • Jumangi - Thursday, September 04, 2014 - link

    These cases are basically pointless. As the article said it could be 2+ years before the case is decided. Then you have possible appeals. The devices that are implicated will be long gone by then. The only beneficiaries will be all the billable hours racked up by the lawyers. Reply
  • smorebuds - Friday, September 05, 2014 - link

    So in your scenario, it's pointless that within 2 years Samsung/Qualcomm would have to come up with a non-infringing gpu design? The more likely outcome (if Nvidia wins/Samsung settles) is those devices will not change much, but would have licensing fee's attached to the gpu's in them. Reply
  • hung2900 - Thursday, September 04, 2014 - link

    I thought that those Apple's chips also used Imagination's GPU? Reply
  • tuxRoller - Thursday, September 04, 2014 - link

    They most certainly do. Reply
  • name99 - Friday, September 05, 2014 - link

    Apple's chips being sold today do.
    For all we know, the A8 will ship with an Apple designed GPU...
  • tuxRoller - Friday, September 05, 2014 - link

    Today, and every GPU enabled mobile device since the iPhone. Reply
  • Samus - Friday, September 05, 2014 - link

    Apple doesn't use Qualcomm chips. This case is focused on Qualcomm and Samsung devices that use Qualcomm. However, Apple does implement PowerVR GPU's in it's A-series SoC's.

    It is also possible, like many vendors, that Apple actually has a license agreement. I'm waiting to see who nVidia's licensee's actually are.

    This case is clearly a scare tactic by nVidia to inevitably get these companies to either license, or straighten out the mobile GPU landscape to determine if the suppliers, vendors, or OEM's of final products license.
  • dylan522p - Saturday, September 06, 2014 - link

    Apple has a deal with Nvidia. Reply
  • thexile - Thursday, September 04, 2014 - link

    Tegra can't compete thus Nvidia decided to sue. Reply
  • Alexey291 - Friday, September 05, 2014 - link

    Pretty much Reply
  • TheJian - Saturday, September 06, 2014 - link

    You apparently missed all the K1 benchmarks? LOL Beats S805 and S810 won't change it either. So how exactly are they behind at this point? CPU and GPU tops right now, and with Denver they got the nexus. Not bad for a start of your desktop gpus in socs. It's a new era now. With gaming being the #1 thing people do on these devices, everyone else is entering NV's (and AMD's) wheelhouse now, rather than the other way around in a modem competition. AMD needs to enter this race ASAP, as both sides have a HUGE edge in drivers, working with game devs etc that the rest will all have to catch up to. The differentiators going forward in the top devices will be your gaming chops, not your modem chops. We'll see what apple has shortly, but S810 is already beaten and is off the shelf arm IP, so battery and perf won't be like custom kraits have been before. Again, it's a new game for a while at least until Qcom answers AFTER s810, so for now NV rules the gpus handily.

    More devices will fall NV's way after Nov's nexus launch. Increasingly mobile=games, so whoever dominates the games will win the next 5yrs of these devices. There is room for two, which is why AMD needs to get in this before that edge is wasted. Consoles won't save them, they need to be in mobile socs too (servers don't count for much as a soc, there's a reason NV avoided it until the ecosystem is there for arm64).
  • SilthDraeth - Saturday, September 06, 2014 - link

    You do realize that the Adreno GPU where originally developed by ATI as Imageon, and AMD acquired it. Then Qualcom bought it and renamed it to Adreno, an anagram of Radeon.

    I don't know how that works, and if AMD signed an anti compete when they sold it off. But they basically stepped out of the mobile arm gpu race.

    I would like to see them back in, but I am not sure how that would really work out.
  • GC2:CS - Saturday, September 06, 2014 - link

    I heard that nvidia got a deal with samsung, to use K1 in the galaxy note 4.

    Samsung had thrown that out of window as they were unable to put K1 into such small device without melting the plastic case.

    NVidia got angry because that means that K1 failed to win any significant devices and sued samsung for not implementing K1 and Qualcomm for kicking K1 out of note 4.
  • dylan522p - Saturday, September 06, 2014 - link

    Nvida already said K1 was not for phones 8 months ago.... They need to catch up in radio before they try to push back in so they went with a tablet SOC. Of course some dumbass chinese companies will go ahead and put it in a phablet. Reply
  • formulav8 - Wednesday, September 10, 2014 - link

    haha, probably right Reply
  • tuxRoller - Thursday, September 04, 2014 - link

    This seems dangerous.
    I'd Imagine that there are others with very extensive, mobile specific, gpu portfolios.
    Perhaps Qualcomm, or samsung, will just buy nvidia?
  • ArthurG - Thursday, September 04, 2014 - link

    Nvidia is by far the biggest innovator in graphics. Their patent portfolio is unmatched and they fill hundreds of new patents every each. Not the ones like rectangles with round corners, but fundamental stuff (like the old ones) and new highly technical ones. In their field, Nvidia has no match and if they want, they can sue every other companies making GPUs. It's not an hazard if Intel pays Nvidia hundreds of millions in licence each year... Reply
  • tuxRoller - Friday, September 05, 2014 - link

    The intel licensing has other factors.
    Img tech has been the major mobile gpu designer for more than an decade (and in existence since 1994), and make the lowest overhead drivers around. If they don't have patents all over the place I'd be amazed.
    There's something behind this that's not obvious. Apple is the BIGGEST seller of img tech around. Why not go after them?
  • djgandy - Friday, September 05, 2014 - link

    The US patent system is a joke though, hence ImgTec won't have a huge portfolio because in the UK you can't file a patents for such generic concepts or things that are clearly just an evolution

    Some of these patents are valid ideas but in no way unique or innovative. What you see with some of these patents is a company registering an idea which was obvious, but simply not implementable in the late 90's because the chip would have been the size of a desk. Programmable pipeline with instruction set. Who could have thought that up?! Sumarised to, instead of the fixed pipeline with complex groups of operations we have currently, we are going to stick an FPU in the mix and you can use the fixed set of instructions that provides instead.

    The innovation in GPU's is in the implementation details, right deep down. That is why Nvidia can't get their power consumption low enough despite their huge patent portfolio.
  • tuxRoller - Friday, September 05, 2014 - link

    Bitboys did a lot of fundamental work before Nvidia even existed. Attacking a company with the patent portfolio the size of Qualcomm isn't smart. Again, maybe Nvidia is looking to be bought out? Either Samsung or Qualcomm could do so easily. Reply
  • TheJian - Saturday, September 06, 2014 - link

    Qcom doesn't have tons of gpu patents. You're mistaking them for a gpu company when they are a modem company. Note they're in trouble for those too though, and Icera is mentioned in the EU suit against Qcom...LOL. NV owns them now, so another suit at some point I guess...LOL.

    I think Nvidia is looking for a large payout for willful infringement of their tech. It might take a few years, but the bills will be big unless they settle if proven real, and knowing they did it for two years NOT settling, will make the bill larger to any jury. You have to be willing to sell for it to be easy to be bought, as a hostile takeover of a company this size would not be simple. There is no reason to sell if you love what you're doing (are fairly young, as Jen is) and know you have the enemy by the financial balls anyway. He has at least a dozen years+ left in him and seems to love getting on stage, so can easily wait this out.

    Nvidia isn't a sue happy company, so you'd have to assume they believe they'll win. Also, the choice of a jury over a judge says they're looking to score big and think the evidence will give them a big payout. It's too easy to pay off a judge these days, so they avoid that and people love to stick it to a big company like Samsung (35B profits, NOT a USA company either) when given the chance vs. a little guy with cold hard evidence ;)

    But again, modem patents won't save your gpu from litigation and bitboys didn't have a MASSIVE gpu portfolio (they never even got a gpu out the door for desktops). I don't see how they could have many gpu patents being 20yrs late to this game (playing actual games like a PC now, the previous mobile crap never did anything like the last 2yrs) and buying an AMD failure which was bitboys vaporware. But I guess we'll find out in a few years.

    Attacking a company that you know is infringing your patents is VERY smart. I'm sure they are aware of whatever bitboys had at the time and since the team was bought. Also, I'm not sure if ONLY the team was bought, or actual IP too and if so how much. Again we'll see soon. They were only founded 2yrs before NV, so it's not like they lasted long, nor had any huge lead and like I said it led to no product. Their R&D was on early mobile stuff, and most of the real advancements didn't come until the late 90's and beyond for gpus. I doubt their early patents would be relevant to tech that would be infringing today. It also says something that even after AMD bought them they were useless and sold. Qcom obviously made something out of them, but before that like you're talking here, were fairly useless. Bitboys didn't make a dime for 11yrs, then in 2006 bought by ATI for 44mil. Remember ATI went for ~$5.4B in comparison (though only worth 1.8B IMHO, and the 3B write-down proved it), so how much did the have that was worth anything? Again, Qcom is a WIRELESS giant, not a gpu giant. There wasn't much graphical innovations going on in mobile until the last few years (and they're just re-treading the desktops really, nothing new, IE - NV/AMD already been there done that, time to pay one or the other probably). You're making bitboys much bigger than they ever were. Three or four vaporware products isn't something to brag about right?
  • tuxRoller - Saturday, September 06, 2014 - link

    You're forgetting that Qualcomm is a far bigger company than Nvidia, and they could devote more resources to GPU development than Nvidia even without it being a core competency. I don't think they do, but their GPU IP stretches back to almost a decade before Nvidia existed. Lastly, with their massive resources they are focused on a single market: mobile. Nvidia has to cater to all markets, hence their marketing jabber about Maxwell being all things to all people. Saying that is a bit equivalent to saying a CPU is more efficient at signal processing than a DSP. That's a tradeoff that's not going away until there's a major design breakthrough.
    BTW, I'm not saying Qualcomm cares enough about this market to either buy Nvidia out (if the suit is successful and the result sufficiently high then maybe) or start spending more on r&d and hiring, but IMG tech already puts out a more efficient product than anyone right now (including better drivers as judged by overhead).
  • TheJian - Monday, September 08, 2014 - link

    You can read the patents for yourself. The problem for qcom is the were making mobile stuff that probably didn't tread on Nvidia's patents until now that they are doing FAR more complicated stuff. Note all the dates of the patents. They are mostly for stuff that is going on today in desktops, and therein lies the problem for qcom. Their patents (if they have many) are not for doing things like we do today on desktops (and phones/tablets now are doing). You didn't game on a phone 10yrs ago.

    The size of a company means nothing in regards to who owns a patent. You either have the patent or pay the guy who does no matter how small he is. IMG was put in phones by AMD/NV :) Meaning, they couldn't cut it with the desktop world and were driven into mobile. Not sure I agree with your claims anyway but I'm fairly sure they don't own a ton of patents in the areas desktops have been playing for a decade either. You can have all the OLD mobile patents you want, and none of them are involved in playing halflife2 etc right? That's just an example but you get the point (far more complicated than some dots game like supersnake etc, read the patents). The patents being argued over here are displaying today's PC/console games not the crap from the last decade of mobile. Qcom cares about what is evolving in their phone/tablet market, or they'll lose it, as it moves to games being #1. They have no choice. Their gpu ip may strech way back but they had no reason to be developing patents for gpu tech they wouldn't use for a decade or more, that they now obviously need.

    What driver info do you have vs. K1 drivers that shows IMG has better ones?
  • tuxRoller - Thursday, September 11, 2014 - link

    I'm not sure why you posted the patents in question as I'm not expert enough for those to be meaningful to me (that includes knowledge of IMG/Qualcomm patents), and from the content of your posts I suspect you aren't either.
    What you keep ignoring, for some reason, is bitboys. They are included in fundamental work before Nvidia was in existence, and the same is true, to an even greater extent with IMG.
    Img has the most efficient mobile arch out there with their well-known tbdr patent ( I'm not going to argue the history of IMG on the desktop as AT has already recounted it so I'll let you Google it.
    The driver data is cited from basemark driver call overhead. IMG is by far the most efficient(a commenter on the link I provided wondered if the reason for that was their tbdr design; I couldn't say what THE reason is, or even if there is ONE).
    Do you have a list of the graphics patents that Qualcomm owns?
    BTW, if Nvidia is bought out how will that affect your identity as it seems you associate with them as strongly as some folks do with apple?
  • Impulses - Friday, September 05, 2014 - link

    Maybe Apple is already a licensee? Which seems really unlikely given their track record but who knows, maybe they just sued the two big fish that are more likely to come to an agreement first. Who knows, Apple without Jobs is a different animal. Reply
  • Jackson07 - Friday, September 05, 2014 - link

    Given that Nvidia had the courtesy to let Qualcomm and Samsung know about their infringement and to suggest a license agreement, I believe they would have done the same with other companies. Maybe others accepted, or are still in negotiations or even other new possibilities.

    Since Apple and HTC, for example, have their history in the patent crossfire, their experience might have fallen to negotiate/accept the agreement. And maybe a reason for that would be just not to spoil their brand marketing. And for Apple, that would be reason enough to pay I guess.
  • dylan522p - Saturday, September 06, 2014 - link

    Apple and IMG have deals with nvidia.... Reply
  • Penti - Tuesday, September 09, 2014 - link

    Imagination is mentioned as infringing tech in the ITC complaint. So is ARM. Reply
  • squngy - Wednesday, November 19, 2014 - link

    AFAIK apple was an nVidia partner and one of nVidias biggest clients. Reply
  • jwcalla - Thursday, September 04, 2014 - link

    Sounds like an act of desperation, in particular targeting Samsung instead of, say, ARM and PowerVR. Reply
  • beginner99 - Friday, September 05, 2014 - link

    Yes because a previous commenter already said it. If your product (Tegra) isn't competitive, use other tactics than actually making a good product. Maybe the deal will be samsung has to sell phones with Tegra SOC? Who knows... Reply
  • TheJian - Saturday, September 06, 2014 - link

    Arm and IMG.L don't make much money compared to the other two and Qcom/Samsung are the two biggest infringers if the case is real (other than apple, and they either have a lic or will be next on the list to sue once this case is over and used against others). Would you go after the broke guys or the rich guys? IMG.L does't make more than $60mil a year, Qcom pulls down 6B. ARM doesn't make much more than NV in any given year. Samsung pulls down $35Billion.

    You target the people getting rich off your stuff, not the broke ones...LOL. Why squeeze blood from a turnip when you can squeeze the actual blood bags instead?

    Also IMG.L doesn't make chips, just IP they license where Qcom actually makes socs. ARM is mostly an IP company also (and again 1/10 of qcom). So again you sue the people making the product I'd think, which is where all the major profits come from. What is the point in suing a homeless man?...LOL. Samsung and Qcom make both look like homeless people ;)
  • tviceman - Thursday, September 04, 2014 - link

    Everyone who makes / uses a cellular modem pays a licensing fee to Qualcomm, so I don't really see how this would be much different. Insofar as how big Samsung and Qualcomm are vs. Nvidia,
    Intel could have easily just swallowed Nvidia and farted them out over the next 2 years when their litigation was taking place. Instead, they settled. I think Samsung (ultimately Qualcomm) will do the same.
  • lilmoe - Thursday, September 04, 2014 - link

    Why Samsung? Why not the vendors of the respective GPUs?
    If Samsung is guilty of using a CPU they "licensed" from ARM, why aren't they going after everyone using Mali and PowerVR?
  • hahmed330 - Thursday, September 04, 2014 - link

    I thought of something... Maybe in exchange for dropping the lawsuits against Samsung nvidia instead of settling for cash or even a licensing agreement... Samsung decides to use GeForce GPU or even their iteration modem for their exynos processors... Stealing marketshare by ligitation instead of making a quick buck or forcing licensing agreements... Reply
  • saliti - Thursday, September 04, 2014 - link

    I think Samsung had some deal with Qualcomm about only using Snapdragon in their phones/tablets. This means Nvidia will have no chance of getting Tegra into Samsung phones, which means Tegra will fail no matter how good they are. Qualcomm/Amd is also using Samsung fabs. I think Nvidia wants to settle by getting Tegra into Samsung phones. Reply
  • xrror - Thursday, September 04, 2014 - link

    I haven't dug into the patent tree yet, so I must disclaim that the following comment might be entirely baseless. However those with a sense of hardware humo(u)r and history I hope this at least give a bit of a chuckle:

    Bitboys Oy !
  • Broken - Thursday, September 04, 2014 - link

    Yeah, but as far as I can remember, they never put out a product, just stole investors money... Reply
  • aryonoco - Thursday, September 04, 2014 - link

    Going after ARM and the Mali line is going to be difficult for NVIDIA from a relationship point of view. After all NVIDIA licenses both the SoC and the ISA from ARM. ARM could seriously hurt NVIDIA if they were to severe relationships with them.

    Apple poses similar problems. Sure they don't use any NVIDIA technology in the iPhone, but they still ship a good number of NVIDIA GPUs in their Macs. The margins on these GPUs is very high, compared to the money that the Tegra unit makes. So again, it's difficult to sue one of your most important customers.

    NVIDIA has no direct relationship with Qualcomm or Samsung, so they are easy choices to sue. But after that, it gets more difficult. And if NVIDIA is serious about integrating LTE modems into its SoCs with the Icera stuff, well then, expect to pay a handsome money to Qualcomm, as they pretty much own all the patents that can be found in that space!
  • Communism - Friday, September 05, 2014 - link

    Excellent observation! :D Reply
  • jerrylzy - Friday, September 05, 2014 - link

    Nice points! Reply
  • TheJian - Saturday, September 06, 2014 - link

    Qcom just had a lawsuit started against them by EU and Icera is mentioned as the competition they tried to harm. Not sure NV would have to pay a dime, and if the suit goes against Qcom, not sure they wouldn't have to pay NV at some point over the icera situation. The point of buying a modem company is to get out of paying all kinds of patents at least to some extent. They already integrated into T4i though we're still waiting on products with it. It seems they gave up on the low end so it may never see the light of day. I do think we'll see a Denver phone, just not sure if it will have icera in it or not. Denver should drop the power on the cpu side and no doubt they'll optimize the gpu side a big by then also (another 5-6 months of work into that chip past K1). I'd be surprised if we didn't see at least a few high end models with K1 of some variety in them, but not too surprised if it contains another vendors modem with it. The nexus in two months will put people on notice as to who is the gaming king for a while. S805/S810 won't change this, so if you're selling a gaming oriented device you're going NV I'd think. As modems shrink you'll be able to use any gpu with them and it won't be so important to have it integrated if gaming is #1 on users minds. Handy to have them integrated in a junker model to cut costs, but not so much in a $400-700 high end model where a few bucks means nothing.

    But yeah, going after ARM or even IMG.L when there can be no huge payout anyway seems pointless. ARM's last quarter being bad was DUE to the GPU sucking wind according to them. So no point in suing losers who already show signs of not keeping up (ARM I mean, never mind IMG.L who can't crack 100mil profits). As desktop gpus take over everyone will be eating NV (and hopefully soon AMD) dust yearly. The decade of modems is over, now it's the decade of gpus. I can't even watch a 720p movie without hitting a 2GB cap, so party over until everyone is unlimited and I don't see mobile companies doing that anytime soon.

    I don't think NV is worried about pissing off apple though. For the same reason Apple doesn't care about pissing off samsung. They need each other anyway, or AMD would be the only choice and many times that isn't desired especially in workstations for pro apps. NV owns 80%+ of that market for a reason and apple can't ignore that just like samsung can't ignore how many phones apple sells netting them Billions in profits whether they hate them or not you take the money.

    Macs only have 10% market anyway, and it's not a huge part of NV's business in the small amount using NV. Having said that, I think you sue the biggest fish AFTER you seal the case on the smaller one with less interest between you (as you noted, no ties to samsung/qcom yet). Prove the case first on sizable fish that will net some bank, then get the rest to settle or use the proven case on the ones that choose not to settle. Not to mention these two are direct competition (they both make socs, Qcom being #1) and one (samsung) has phones/tablets you want into (#2 in mobile 2nd only to apple profit wise) and fabs that can be negotiated over also. Good choices to sue IMHO assuming you have a case.
  • Khato - Friday, September 05, 2014 - link

    Definitely an interesting move by NVIDIA. One which is likely motivated by Samsung burning them on a product design win in favor of Qualcomm. I also wouldn't be surprised in the least if NVIDIA's competition is being uncooperative/unreasonable with respect to licensing as the mobile landscape seems to prefer 'cold war' type patent stand offs rather than the reasonable cooperation that we're used to in the PC space. Which is why I fully expect Qualcomm to fire back at NVIDIA on the modem side somehow - aka I'd guess that NVIDIA was looking to enter into a cross-licensing agreement with Qualcomm that would basically boil down to graphics for modem. Reply
  • aryonoco - Friday, September 05, 2014 - link

    Graphics for modem seems like fair on the surface, but isn't Adreno basically the old ATI Imageon team that Qualcomm purchased from AMD (for cheap)? IIRC, Qualcomm also got a good package of patents with their "ATI handheld" purchase. And QC has been filing for patents left, right and centre in recent years. So, the issue is more complex than might seem at hand.

    Anyway, it will all work out in the end, and I very much doubt that they this will even go to trial. They all have just sit down and work out how much to pay each other, and then it's done. NVIDIA has basically now forced those negotiations to happen, that's it.
  • tuxRoller - Friday, September 05, 2014 - link

    Those guys were real pioneers. The number of patents they held must be impressive.
  • CoreDuo - Friday, September 05, 2014 - link

    I love this line in that article: "I'm calling it right here: three years from now, Intel will overtake Qualcomm in the MSoC business. Discuss."

    If Intel is going to overtake Qualcomm, they have about four months to do it according to the author.
  • tuxRoller - Friday, September 05, 2014 - link

    I noticed that as well. How often do you hear analysts being called on their poor predictions?
  • Khato - Friday, September 05, 2014 - link

    Oh, I'm certain that Qualcomm got a number of patents along with that deal, and they certainly have been filing patents since the purchase... But the AMD-NVIDIA cross-licensing agreement would no longer apply to the sold division necessarily and hence Qualcomm is left wide open. I'd imagine that NVIDIA was fine with this initially, but as Qualcomm gained prevalence and NVIDIA wanted to enter the market? Yeah, then it'd be a good chip to use for leverage, but why would Qualcomm play ball when they're under the impression that their modem IP is far more valuable than the graphics IP that everyone else just cross-licenses?

    Of course that's just pure speculation, but it wouldn't surprise me in the least if that's what we end up with when the dust settles. Which I completely agree will likely be without an actual court trial, well, at least on the Qualcomm side. NVIDIA could be adequately annoyed at Samsung's design practices to try and hold their feet to the fire with the goal of either getting into Samsung products or aiding the competition which just might be using their products.
  • saliti - Friday, September 05, 2014 - link

    It was rumored that Samsung got heavy discounts on Snapdragon for the Note 4. Maybe Nvidia was in the running for getting their product in Note 4 ? Reply
  • Khato - Friday, September 05, 2014 - link

    Given what I've heard of Samsung's mobile product development practices I'd be rather surprised if Samsung didn't have a Note 4 version using a Tegra SoC designed and prototyped. It's quite possible that they even would have had a Merrifield/Moorefield version. Apparently Samsung makes a habit of designing multiple variations up through the prototype stage, which is a great strategy in order to get the most favorable pricing - the more I learn about Samsung the more I dislike them. Reply
  • Penti - Friday, September 05, 2014 - link

    Actually they contacted Samsung in august 2012, not Qualcomm and in the complaint assumes Samsung later spoke to Qualcomm about it.

    From (28) page 9-10 plus (34) page 11 in the Delaware complaint.

    "Since August 2012, NVIDIA has attempted to reach an appropriate license with Samsung, which would enable Samsung to properly use NVIDIA’s IP within its products. But Samsung has negotiated based on delay and by pointing the infringement finger at its chipset suppliers, such as Qualcomm, or third parties that supply GPU technologies used by Samsung in its own processors." + "Qualcomm and Samsung have had knowledge of some or all of the Asserted Patents since before this Complaint was filed. Samsung knew of at least the ʼ488, ʼ667, and ʼ063
    Patents since at least August 7, 2012, and at least the ʼ685 and ʼ913 patents since at least January 15, 2014, through discussions with NVIDIA. Upon information and belief, Qualcomm knew of the same patents at least through discussions with Samsung." Thats some 5+ years after PowerVR SGX was sampled plus Nvidia had no real products during 2009-2010 or early 11 when the market was really taking shape and offered no OpenGL ES 2.0 GPU (mobile SoC) until Tegra 2 with an appropriate CPU, which still lacked NEON. Plus Tegra APX was later to the market than solutions like Samsungs APL0298C05, TI's OMAP3430 and so on. It's some ~4 years after Qualcomm announced that they bought the Imageon line. It's 5 years when it comes to '685 and '913. Patents that were approved during 2001-2007 (applied for between 97-2003) long before Qualcomm started making their own designs. It seems Nvidia hasn't made an effort or at least claims they haven't spoken to Qualcomm despite the fact that Nvidia needs to license lots of 3G and 4G related stuff from them as well as gpu technologies. Which Qualcomm has patented by employing people working in the field even before they bought was to become the Adreno-gpus. So I don't question Samsung's reaction that it is a matter for the suppliers, not something you bring up half a decade later to a decade to the assembler of the components. After all some of the patents were published and approved way back in 2001. AMD, Qualcomm and Nvidia should have made an agreement back in 2008 when Qualcomm and AMD made their agreement to transfer the unit. Not hearing it from a Qualcomm customer second hand in late 2012 or so. It does look like Nvidia made no efforts before this time, but of course they did have knowledge of Imagination's, Vivante's, ARM's and Qualcomm's gpu businesses and did have patent license agreements (as well as other IP) with these companies for different technologies.
  • testbug00 - Friday, September 05, 2014 - link

    So, why not sue Apple? Apple has more given it uses PowerVR... only reason I can think of not to sue them is that they would lose.

    Seems like higher-end patent trolling,, MAYBE. I'm not sure what cross-licensing ATI/AMD and Nvidia have... iirc, the first unified shader part was the xenox GPU in the Xbox 360...

    Seems full of holes when they say "these companies all infringe" and don't sue them all, or even all the rich ones...
  • Reflex - Friday, September 05, 2014 - link

    As pointed out by others, Apple buys a lot of nV gpus for their Mac line. Chances are high this issue is already part of their existing agreements. Reply
  • testbug00 - Friday, September 05, 2014 - link

    It is? Don't you think Nvidia would shout to the world that they have a licensing agreement with Apple of all companies?

    Also, as for ARM... The company that licenses to almost everyone? Why would Nvidia claim companies infringe on their patents if they are licensing? Seems like a threatening move at most... Or a move of "if we sue them we'll lose" ._.

    Trying to do a full Jury against Samsung makes me think they will try to use the fact that Apple beat Samsung in jury trials to an advantage.
  • Impulses - Friday, September 05, 2014 - link

    They aren't legally required to sue them all or prove their case against all, and there's probably strategic reasons (alluded to in other comments) why they chose those two. Reply
  • jjj - Friday, September 05, 2014 - link

    Not so sure it's hard to decide who should pay , Qualcomm gets payed (a lot) for 3G/4G , this is not all that different. Reply
  • legokangpalla - Friday, September 05, 2014 - link

    Aren't some of the patents mentioned by nVidia on the lines of...patenting standard logic gates like AND NAND etc? If those technologies are so widely perpetuated and are basics of modern GPUs, isn't it a bit push to patent these technologies? Reply
  • Malih - Saturday, September 06, 2014 - link

    Wouldn't this be good news for AMD, strengthen their relation with SAMSUNG and other companies being sued. Reply
  • testbug00 - Saturday, September 06, 2014 - link

    especially if AMD can license them patents that cover it, which, they would likely do for FAR LOWER costs than Nvidia would. Reply

Log in

Don't have an account? Sign up now