The entire ruling is available on NVIDIA's site, and is consistent with NVIDIA's claims of a "favorable ruling". Plus it's a pre-trial hearing, so there wasn't anyone sitting in on this to report it. The real fireworks don't start until the full trial starts.
Short answer is because they need to sue the manufacturer and work up from there. Otherwise ARM would likely have the case dismissed on account of the fact that they don't actually make any of the infringing hardware.
In fact a big component of the case will be who is responsible for infringement. Samsung (who makes the phones), Qualcomm (who makes the chips), or other parties who make the IP.
I think Nvidia will use TSMC's 16nm not Samsung/GF 14 nm if I'm not mistaken. So I doubt Nvidia will care about that, that's why they targeted Samsung and not Apple which would have big leverage over Nvidia.
What's really bogus is Samsung's "counter-suit" is stationed in Velocity Micro's district instead of NVidia's (even though VM is not the primary target) because Samsung wanted it that way just to inconvenience NVidia further.
Huge dick move by Samsung getting VM involved in the first place, and even huger dick move making NVidia fly their legal team across the country countless times throughout the trial.
The more difficult you can make things for the opposing party the better. The more exhausted they become, the more irritated they become, the more expensive it is for them, the better.
Smart move on their part. If at any time you can get a favorable venue, you should elect to do so.
What makes you think Apple is immune from nVidia's lawsuit? It's much easier to score a victory against a foreign company first -- they are less likely to win in a jury trial -- than to go head to head against another domestic company.
@samus : it's called secondary liability. That's also why nVidia doesn't have to go after another American company like Qualcomm and has chosen to go after the customer of Qualcomm.
Sure, they're patents could be valid. The fees they asked for? Knowing Nvidia it was probably insanely high.
Based on my estimates, (from: http://www.anandtech.com/show/7112/the-arm-diaries... ARM would at most take something like 2 dollars on a $40 chip. assuming it used all ARM IP and an ARM layout. That's a high estimate for ARMv8. More likely it would be closer to 1.20-1.25 USD. That's for a high end SoC. Most SoCs are likely 20 dollars or less. So, 50 cents to a dollar goes to ARM. That's for a fully ARM designed chip.
My guess is Nvidia is probably asking more for their GPU than ARM would for a whole chip including the ARM layout for the foundry and note you want.
However, the could be partially due to wanting the court to lower that to a about market average fee instead of sub-market-value or something like that. Time will tell.
Quote: Sure, they're patents could be valid. The fees they asked for? Knowing Nvidia it was probably insanely high.
As the head AMD troll from the S|A site I realize that you would never ever read or research any facts regarding Nvidia so here is the actual wording from Nvidia's suit that puts you're biased FUD back where it belongs, in the dumpster where you found it.
Quote: We are asking the courts to determine infringement of NVIDIA's GPU patents by all graphics architectures used in Samsung's mobile products and to establish their licensing value."
Nvidia is a company that makes a lot of money, partially because they are to aggressive at aiming for and achieving high margins. They tend to not make very many "friends" outside of companies that make their GPUs. Microsoft, Sony, countless tablet/phone OEMS... They used Nvidia a few times, than dropped them.
For Microsoft and Sony, it surely had to do with cost for the same performance. Especially given how price sensitive the console market is. For phone/tablet OEMS... Given Nvidia's power and perfomance numbers are true, they must have priced their chips far to expensive. Given there chips were a price to other high end SoCs, they must have had bad support/power/performance/etc.
Nvidia has a history of expensive pricing, I wouldn't be shocked if this is another case. As for "AMD troll" you're welcome to make statements like that. While I try to be objective as much as possible, I have a more positive view of how AMD interacts with consumers and buisness partners than Nvidia.
However, where was AMD part of NVidia v. Samsung/Qualcomm? Perhaps you want to relate the sale of Adreno is 2009? /meh
My first commend would be "hindsight is a bitch" to Microsoft and Sony going with AMD because now that Maxwell is out, a PS4/XBOX One refresh with a more efficient\equal performance part would be tremendous (such as when Microsoft revised the Xenon/Jasper XBOX 360's)
My second comment is NVidia would never sue AMD (or vice versa) because they have a bundled licensing agreement. The purpose of this condensed license is to prevent lawsuits because they essentially cover entire categories of patents, not specific patents themselves. It's ironic the Chinese government fined Qualcomm into a 1 BILLION settlement because the Chinese only believe in "a-la-cart" licensing, when it is well accepted to bundle patents throughout the world to prevent exactly what the Chinese government supposedly fined Qualcomm for.
You make it sound like either company wanted to work with Nvidia. That's the problem. Nvidia screwed Microsoft on shrinks and charging the same price for the original Xbox. Nvidia failed to deliver the performance at the price they promised Sony with the PS3.
Maybe it's just me, but, if I had a one in 8-10 year design, I would want to make sure everything went right. Gamecube, Wii, Wii U, Xbox 360. All ATI/AMD, all fine.
Anyhow, what does this have to do with what I was saying 0.o?! Thanks :)
It just sounds like Nintendo knows how to negotiate a contract. And XBOX 360 was a disaster, it was inches from being the first recalled gaming console in history; 1 die shrink, 2 revisions, 3 years later and 40 watts of power reduction helped to resolve the same issues NVidia was blamed for just a few years prior with "Bumpgate" while Microsoft took all the flak.
I'm not an NVidia fanboy (although I personally find it ridiculous to not consider a Maxwell card these days) but I am NOT a Samsung fan for dozens of reasons. They don't stand behind their products, and I have zero respect for companies like that. Microsoft, Intel, AMD, Nvidia, Apple, even HP Enterprise (with the very annoying exception of out-of-warranty BIOS updates) all stand behind their products. Samsung, Lenovo, Google, Logitech, to name a few, do not stand behind their products: slow to act on resolving customer complaints, denying warranty coverage for bogus reasons, discontinue software support for products often still sold and/or in-warranty, etc.
So yes, for unrelated reasons, I am rooting for the company I know has a good track record of customer support and listening to customers, opposed to the company that I feel is complete opposite. As a (ex-) customer of Samsung I got nowhere with them, so perhaps billions in damages and legal pressure will signal a wake-up call for them to change their shit around.
On a side note, Nvidia sued them FOR NOT LICENSING their products. "NVIDIA’s claim notes that the company has been attempting to reach a license agreement with Samsung and Qualcomm since 2012, and that today’s suit is a result of their inability to come to an agreement over the last 2 years." (http://www.anandtech.com/show/8492/nvidia-files-pa...
Nvidia either offered a (multiple?) costs to license their Kepler GPU and/or patents. They couldn't come to an agreement. You don't go into these things say "how much do you think is fair to pay" unless you have a weak position. I believe Nvidia thinks it has a strong position (correct or not) and hence would demand a large amount. Probably larger than they expected to get to make the other side feel like they "won" a concession. Simple negotiating tactics. =]
It is also believe (but not confirmed) that Intel (and thus Apple, who was at the center of the case) signed a GPU licensing agreement after the 1.5 BILLION dollar settlement in 2011. The fact NVidia isn't going after Intel (or Apple) lends further weight to those license rumors...
They signed an agreement for cross-license for ALL of Nvidia's graphics patents at the time of signing. And some of Intel's patents also to Nvidia.
They didn't license the 7 patents in this case specifically. Intel still should be covered by that license.
And, as for other companies. Just because they're only sueing 2 publically doesn't mean they're going after others. If they aren't going after Imagination Tech, Apple, and any sizable company that uses MALI GPUs, I would be shocked. They might not be sueing them yet, but, certainly have or are in some form of (cross)-licensing talks.
Considering Samsung has been in hot water dozens of times for not licensing patents, and other companies you've mentioned haven't, my guess would be the other companies, as you've suggested, are licensing the patents.
And you're right, Samsung probably looked at it like this: we sold 60 million devices that infringe on this IP during the period they want us to license for (let's just say...) $15 per device sold.
That's almost $1 BILLION right there. Samsungs' legal dept probably said "let them sue us and we'll negotiate it down to less than that." Samsung pays less, legal department stays funded, nVidia gets something, everyone wins.
And you can bet this trial will end in a settlement.
Good seeing Nvidia aggressively pursuing payment from infringing Samsung and Qualcomm, why should they not pay for use of Nvidia's market leading graphics IP when Intel does.
The issue isn't if they pay or not. The question is how much is Nvidia asking for, and, was it willing to accept industry standard rates? The ARM rate for a full GPU designed for you to implement is under a percentage.
I find it hard to believe Nvidia would accept sub 1% royalty. For better or worse.
Why would ARM be any guide to industry standard when its ARM that's infringing on Nvidia's IP.
Nvidia payed for the R&D and patents and ARM takes want it wants and charges Samsung/Qualcomm want it wants this in no way sets the price for a license agreement for Nvidia's IP who owns the IP in the first place. Samsung/Qualcomm/ARM need to negotiate a agreement and its now going to be more expensive because of damages and court costs.
We'll see how this plays out, for better or worse depending on whose side you're on.
Some of the IP companies has made GPU's with the supposedly infringing IP since before some of the patents where filed. Nvidia basically claims that everything other than their chips and Intel's is infringing. Samsung, Qualcomm, ARM, Broadcom, Imagination and others own a lot of GPU IP that is valid too.
"In what NVIDIA calls a “favorable ruling” from the ITC, NVIDIA has received what they consider to be favorable definitions in 6 of their 7 disputed claims."
NVIDIA tried for years to come to some kind of agreement, many did such as Intel but ARM Samsung, Qualcomm and others did not and will be held accountable for infringing on NVIDIA'S Patent. NVIDIA is not disputing that, "others own a lot of GPU IP" NVIDIA is protecting their patents from those who refused to come to the table and hammer out a agreement. If/When NVIDIA wins this case others will come to table and talk but it would of been better for all and less expensive to have talked before it had to be taken to court. If/When NVIDIA wins they will be negotiating on the side of strength.
Intel agreement stems from their 2004 chipset licensing deal with Nvidia. They never tried to block the sale of Intel products or turned to ITC in their Intel dispute. They don't mention any deal with anyone else.
They didn't try for years with Samsung as they didn't actually say which patents (in full, they mentioned three to them on 7 August 2012 and two more on 15 January 2014, they guess Qualcomm knew from discussion with Samsung but didn't talk to them) they wanted to be payed for until January 2014 and didn't talk to Qualcomm at all! Don't be ridiculous just read their argument in their suit instead.
And this is where the actual relevant patents come into play, lay them out on the table for all to see. You don't go from tile-based 2D renders to fully programmable shader architectures overnight, and you can't say "well we just copied everyone else just because".
Its not disputed that Nvidia invented most, if not all of the patents that are in dispute, they have some cross-licensing with ATI/AMD in that regard but the fact the dGPU race has been a 2 horse race for the last decade long before the advent of smartphones and SoCs should tell you the likes of Imagination and ARM probably don't have a lot of relevant IP to make a case....
Imagination made their first 3D GPU back in 95 and Qualcomm bought Adreno from AMD and made lots of contributions in the OGLES2.0 era when Nvidia was absent as well as hiring lots of people and patenting graphics stuff even before they bought the Imageon line. Both Imagination and AMD has been in the game longer than Nvidia. Nvidia didn't go after Qualcomm after they bought the Adreno technologies and as they filed the suit against them they still hadn't talked to Qualcomm according to it. If they wanted to claim that they have the sole right to produce GPU's or sell GPU IP they should have done so back then around 08. They didn't dig up the patents before 7 Aug 2012 and 15 Jan 2014 and they were irrelevant when they finally agreed with Intel in 2011. Most of them patents stuff that was already publicly disclosed or on the market, which you can't do so most claims probably could be invalidated. Samsung and Qualcomm has lots of semiconductor IP which is infringed upon by Nvidia. It's just that nobody licenses every bullshit patent and mostly they are not offered to be licensed either before suits happens. When you license IP you also get sublicensing rights to the technology in many cases and even ARM had fully programmable GPU's in the mobile space before Nvidia.
@Penti, you're wrong, PowerVR was always behind Nvidia and even ATI as their last commercial dGPU card the Kyro2 lacked major functionality such as Hardware T&L. By the time programmable shaders were mainstream (invented by Nvidia btw, without dispute with GeForce 3) PowerVR wasn't even a player in discrete graphics. They managed to survive in embedded designs, then suddenly start licensing IP and enjoying a resurgence in smartphones and smartdevices. Lucky coincidence!!! Unfortunately for them there is a huge gap in what they support and what their IP is capable of, so it should be very easy for the courts to sort that out.
Qualcomm was named in the suit btw, and while they did buy Adreno from AMD it is becoming more and more clear they only bought their engineering team and not any of the actual IP. The fact AMD is talking about getting back into mobile graphics and is also rumored to be licensing their IP all but confirms this.
So, this again leads to the fact Nvidia has a very strong case, after all they have laid out their patents on the table for all to see and have products unbroken in timeline from their inception until now demonstrating their leadership in the industry.
T&L was known before the patent in this suit was filed, S3 had their HW T&L card on the market before Nvidia sought protection ('488) and you can't patent things that are already disclosed to the public or used by someone else, though S3's implementation was broken. It probably takes more than implementing a software pipeline in hardware to actually invent stuff that's patentable too. The scope is probably too broad.
Qualcomm is sued, but they (Nvidia) admit in the filing/suit that they never ever talked to Qualcomm and that they never told Qualcomm that they were infringing or what they infringed upon. Just read the suit as said. That they weren't bothered by Intel's XScale mobile processors with integrated graphics between say 2004-2006 before the sale to Marvell (or concerned about Marvell after the sell i.e. 2006-today) is not going to help. That they weren't concerned about Samsung's ImgTec use between say 2004-2011 (they licensed the tech in 2004 and onwards) is not going to help their case. That they weren't concerned about Freescale's use of PowerVR and Vivante graphics IP between 2005-today is not going to help. Neither should that they never went after TI and others for the use of PowerVR, Vivante or Mali graphics. Neither should it help that they never talked to Qualcomm despite the fact that they had a modern mobile gpu on the market before Nvidia, despite that they knew about the 2008/09 sale of the Adreno tech they never actually informed Qualcomm until they filed the lawsuit on September 4th 2014. In the future it looks like Samsung will be a minor Qualcomm customer and corporations like Microsoft, Xiaomi, Lenovo/Motorola Mobility, Huawei, LG, Sony and others is the main Qualcomm users on the market. Many of the patents hadn't been issued when Intel and Nvidia entered into their 2004 agreement – which was about chipset licensing http://www.nvidia.com/object/IO_17070.html. They also dug up a 3dfx patent to use in the suit.
your licensing estimates are waaaay off. Arm charges way more than that.
Arm charges licensing fee's per Core. Big.Little design's means 8 cores.
The fee per core is determined by device retail cost. on a $650 phone it's more like $15 just for the cores, and another $5 for graphics ip.
The Nvidia vs Samsung case comes down to this.
Samsung was paying Nvidia licensing money for several of their patents. They been paying them for 4 years...
Then when the S5 came out, is when Samsung said, "nah, those patents are to general, we decided we shouldn't be paying for them anymore." So they stopped paying, hence nvidia gets mad and threatened to sue. Which they made good on their threat eventually.
Now nvidia's already mad, and if they are going to go through with all the trouble of a trail, then we might as well be damn sure we go after the right person. So they start at the device maker, and work their way up.
TLDR: Samsung stop paying their licensing fee's and nvidia sued to make them pay.
Nvidia didn't bundle and sell these patents to anyone including Intel which they claim is covered by their 2011 agreement. Nvidia didn't even tell Qualcomm about the patents and only disclosed some of the patents to Samsung in 2014 shortly before they filed their suit. You can clearly see that in Nvidia's filing. They just dug up obscure stuff and didn't even ask for licensing fees until Aug 2012 but without disclosing what they thought was infringing or most of the patents used in the suit.
Seriously...when the S5 came out? Then why does this case represent failed negotiations stemming from Feb 2012 (GS3-era?)
I guess in your world nVidia KNEW Samsung was going to infringe to they preemptively attempted to license? This case may cover half a decade of Samsung devices for all we know...if nVidia began approaching Samsung in 2012, it could very well involve IP used in the Galaxy Tab and the Epic 4G Touch. We'll know more when the trial starts.
Regardless of the merits of the suit, nV is under no obligation to license its patents at all, or to license them at a price you (or Samsung) find "reasonable".
IF you ask for your patents to be made part of a spec (eg WiFi, ethernet, LTE) then you have to offer fair and reasonable licensing. But not in general.
If you want to go down this path of defense, you'd have to show something like nV offered these particular patents to Khronos as a more-or-less essential part of one or more of their specs. I'm unaware of anything like that.
Apple uses Imagination's GPU in every iPhone and iPad, and if nVIDIA wins, much more money it gains, but nVIDIA didn't sue Apple. The reason is simple: this kind of patent sueing is broken in the US. Companies like Samsung is more vulnerable in a case like this, while nVIDIA can't touch Apple in the US (remember: Obama vetoed, lol)
WRONG. They'll sue as soon as they have precedent set from the first case, or use that case to get apple to settle (which if rumors are true apple is already talking to them, and surely after hearing the court reads the patents NV's way so far). Why would you go after the most influential company first? Samsung is outside usa, and Qcom is your mortal enemy at this point with nothing to gain from them (samsung has a modem, fabs, screens etc, so lots to deal over there).
Imagination's PowerVR architecture is named also as something they wanted blocked. One you win this, apple can't ship a phone/tablet either, so no point in hitting them until you shutdown samsung based on the same gpus. Apple won't fight in court if they know they have a total loser, they'll wisely deal as opposed to having to possibly delay a launch that would cost them a huge portion of the market to someone else (probably samsung, once suit is done and settled they'll still be shipping). Apple is likely dealing as we speak to switch to NV gpus (there is no way imagination would win a suit vs. NV, and they make so little <60mil/yr they'd die before it was over), since a victory in samsung suit means the next case would probably have a higher chance of banning products as IMG.L stuff would already be infringing via samsung suit. A court would look and see they already are infringing here "apple" so stop shipping or deal outside court immediately. I'm sure apple would like terms NOW before NV wins and has a sledgehammer in their hands for leverage to use in negotiations.
I don't think anyone would want to be stuck with a diminishing gpu return vs. Intel's side. AMD looks like they're on their way down on gpu just like cpu (can't sell current stuff, so delay on 300's for months, no profits etc), and the apus will get squeezed to death by INTEL vs. ARM war in the race down/up for both. You couldn't go AMD ONLY for long without screwing your mac/macbook etc sales. IE, the current Mac Pro has two FirePro D700's (same as W9000's) that put out 7TF, which is pretty much exactly what NV does with TitanX with far less power used (same story with M6000 probably).
http://hothardware.com/reviews/Workstation-War-AMD... Cuda 3x faster in Adobe Premiere than AMD w/OpenCL. They pretty much take every test here but sony vegas (who uses that instead of premiere when buying CUDA cards? Only Anandtech...LOL). Check out the conclusion page on how Cuda is massively used by apps and OpenCL in some cases is just dumped (pages on sites not updated since 2012...LOL).
Anandtech chooses sony vegas because you don't want people knowing Cuda is 3x faster ;) Considering the tests results of all the benchmarks they ran (and it was vs. Kepler/fermi), you can't screw NV without screwing yourself :) It boggles the mind when I see anandtech avoid cuda like the plague vs. AMD in anything and OpenCL. They should choose the most popular apps that use both, or pit say, Sony Vegas+AMD/OpenCL running the same test/operations vs. Adobe Premiere+NV/Cuda. There are a ton of apps you can run OpenCL plugins vs. Cuda versions (3dsmax, Premiere goes either way in mercury engine, Cinema4d, Blender, Lightwave, Maya etc). Yet Anandtech keeps running stupid synthetic junk or stuff nobody makes money with (bitcoin mining, F@H etc...LOL). I digress...
One quote from the last page of the article that pretty much sums it all up: "If you want to use V-Ray, i-ray, or Octane Render, you effectively need an NVIDIA GPU -- V-ray might technically support OpenCL, but we couldn't get it working. The company's OpenCL page hasn't been updated since 2012 and basically tells users to switch to CUDA. Other applications, like SolidWorks and AutoCAD, tilt so strongly in Nvidia's favor they may actually offset the increased price." (note maya/creo in spec favors NV now after NV fixed drivers that caused a regression in perf). For pros, price usually isn't an issue and NV pretty much wins period.
Exactly, I think Nvidia went after Samsung/Qualcomm first to set precedence, but also because I think they are already working with Apple to license their GPU IP. It doesn't make sense to go after Apple if they are already in talks to license and I am sure the weight of those license fees will depend largely on the outcome or probability of a favorable outcome from this initial IP case.
Yay Nvidia wins, AMD close shop. :) I really hope this is the case because that way I get to see Nvidia fan boys pay $9999 for a Geforce 205, or $12000 for a 50Mhz overclocked one.
Some idiot was saying this 9 years ago when Intel Conroe'd AMD...we'd be paying $10000 for our next CPU etc. etc.
Oh wait, none of that ever happened.
CPUs, GPUs are luxury toys, that's it, as such there's a ceiling on the price people will spend on them limited to their discretionary income. Intel, Nvidia, Samsung, Apple, they all know this. They can only charge so much for luxury toys, if they charge too much they just start hurting themselves by reducing total sales.
How do you know that Imagination or Apple haven't ALREADY licensed from nV (or even that they don't infringe)? Neither have a history of blatantly ignoring patents. You're constructing a narrative based on how you want the world to be, not on any actual facts.
Compare, as an example the lawsuit filed after the release of Cyclone about memory-address disambiguation. That has disappeared without a trace; presumably because either Apple licensed the patent, or there was no infringement as soon as Apple explained what they were doing. (This is not so strange. When I was at Apple, we were sued by a large patent holder who was convinced that we were infringing some patent they had on how to connected decoded MPEG video frames to an underlying time base. Their lawyers explained the patent and why they thought we had to be infringing, and I told them that's very nice, and the naive way they described of solving the problem works, but it kinda sucks for the following reasons, and so therefore I used this much more sophisticated alternative. Point is, it was all a quite reasonable technical discussion: "we think you have to be doing A to solve this problem", "you're wrong, we do B to solve the problem", "ahh, you're right, NICE! OK, sorry to have bothered you." )
If the license was strong enough that Apple was licensing wouldn't Nvidia have said something? Did the license agreement with Apple state it cannot be disclosed? If so, wouldn't that imply that Apple would get something out of it? A free license? A super cheap license? It seems Nvidia would have more to gain having Apple out in the open besides Intel (who never specifically licensed there patents, making them a very weak example (imho, of course)) especially given Apple's willingness to fight over patents.
That would indicate the patents aren't worth very much. Doesn't make much logical sense to me. As always, I could be wrong. It seems having an Apple license and keeping quite about it is a lose-lose no matter how it happens.
I suspect this just means they'll get a better deal when all parties come to a settlement. And I bet nVidia's primary motivation in this case is to compel Samsung to rely less on Qualcomm and more on them.
I also expect they want a payout from Qualcomm to grease the wheels.
This is a very complex case at its core, dealing with: 1) who is responsible for licensing IP? Chip designer, chip manufacturer or device assembler (in this case, ARM/Qualcomm, Qualcomm/Samsung and Samsung respectively) 2) are mobile processors infringing nvidia IP? This will be settled in court, but signes point to yes, as the biggest part of the desktop graphic race (and as such research) was done by the 2 horse race of nvidia and AMD. They have a cross licensing agreement, but it seems Qualcomm did not inherit from that during their purchase of Adreno. 3) What remedy will be available? Does nvidia want licensing and payment or just ban all their competition (they would be in their right to do it if they won, as they have no obligation to license those patents).
A case that will decide the fate of mobile 3D and probably fatten up nvidia's pockets.
If you read the suit you see that they never talked to Qualcomm and the sell of the Adreno tech would not have happened if there were any intellectual property problems. Nvidia certainly use semi tech developed by Qualcomm, Samsung and ARM, and even Imagination does have a range of graphics patents which they infringe. Going to ITC makes it all look bad for Nvidia as they claim customers can happily buy Intel or Nvidia chips instead. But Nvidia was mostly absent in the early 2007-2010 era and didn't have any products (at all) to sell when the smart phone platforms (OS's) took form. Neither did Intel in that specific era and Intel was in dispute with Nvidia over cross licensing until 2011. First 3 patents were disclosed to Samsung on 7 August 2012 and the last 2 on 15 January 2014. So they can't really claim that the supposedly infringing products could have been replaced by something else, but they do try. The patents weren't used or disclosed previously and they are not really any strong patents. They might get cross licensing deals out of it, but I couldn't imagine why anyone would pay them for that. It's not like this tech was developed in secret but that is required to patent it.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
53 Comments
Back to Article
Shahnewaz - Monday, April 6, 2015 - link
Source: NvidiaOh come on!
Guess I have to find a report from someone not involved in this lawsuit.
Ryan Smith - Monday, April 6, 2015 - link
The entire ruling is available on NVIDIA's site, and is consistent with NVIDIA's claims of a "favorable ruling". Plus it's a pre-trial hearing, so there wasn't anyone sitting in on this to report it. The real fireworks don't start until the full trial starts.lilmoe - Monday, April 6, 2015 - link
I guess that means no 14nm for NVidia.Jokes aside, I'm not sure why they're going after Samsung in the first place and not after ARM...
Ryan Smith - Monday, April 6, 2015 - link
Short answer is because they need to sue the manufacturer and work up from there. Otherwise ARM would likely have the case dismissed on account of the fact that they don't actually make any of the infringing hardware.In fact a big component of the case will be who is responsible for infringement. Samsung (who makes the phones), Qualcomm (who makes the chips), or other parties who make the IP.
vladx - Monday, April 6, 2015 - link
I think Nvidia will use TSMC's 16nm not Samsung/GF 14 nm if I'm not mistaken. So I doubt Nvidia will care about that, that's why they targeted Samsung and not Apple which would have big leverage over Nvidia.Good luck Nvidia, hope you'll win!
Samus - Monday, April 6, 2015 - link
What's really bogus is Samsung's "counter-suit" is stationed in Velocity Micro's district instead of NVidia's (even though VM is not the primary target) because Samsung wanted it that way just to inconvenience NVidia further.Huge dick move by Samsung getting VM involved in the first place, and even huger dick move making NVidia fly their legal team across the country countless times throughout the trial.
looncraz - Monday, April 6, 2015 - link
Tactics, tactics, tactics.The more difficult you can make things for the opposing party the better. The more exhausted they become, the more irritated they become, the more expensive it is for them, the better.
Smart move on their part. If at any time you can get a favorable venue, you should elect to do so.
Samus - Tuesday, April 7, 2015 - link
I agree, but neither venue is favorable to Samsung. However, I guess California is "less" favorable given Samsungs' experience there.shm224 - Sunday, April 12, 2015 - link
What makes you think Apple is immune from nVidia's lawsuit? It's much easier to score a victory against a foreign company first -- they are less likely to win in a jury trial -- than to go head to head against another domestic company.shm224 - Sunday, April 12, 2015 - link
@samus : it's called secondary liability. That's also why nVidia doesn't have to go after another American company like Qualcomm and has chosen to go after the customer of Qualcomm.testbug00 - Monday, April 6, 2015 - link
Sure, they're patents could be valid. The fees they asked for? Knowing Nvidia it was probably insanely high.Based on my estimates, (from: http://www.anandtech.com/show/7112/the-arm-diaries... ARM would at most take something like 2 dollars on a $40 chip. assuming it used all ARM IP and an ARM layout. That's a high estimate for ARMv8. More likely it would be closer to 1.20-1.25 USD. That's for a high end SoC. Most SoCs are likely 20 dollars or less. So, 50 cents to a dollar goes to ARM. That's for a fully ARM designed chip.
My guess is Nvidia is probably asking more for their GPU than ARM would for a whole chip including the ARM layout for the foundry and note you want.
However, the could be partially due to wanting the court to lower that to a about market average fee instead of sub-market-value or something like that. Time will tell.
HighTech4US - Monday, April 6, 2015 - link
Quote: Sure, they're patents could be valid. The fees they asked for? Knowing Nvidia it was probably insanely high.As the head AMD troll from the S|A site I realize that you would never ever read or research any facts regarding Nvidia so here is the actual wording from Nvidia's suit that puts you're biased FUD back where it belongs, in the dumpster where you found it.
Quote: We are asking the courts to determine infringement of NVIDIA's GPU patents by all graphics architectures used in Samsung's mobile products and to establish their licensing value."
http://techreport.com/news/27011/nvidia-tries-to-b...
testbug00 - Monday, April 6, 2015 - link
Nvidia is a company that makes a lot of money, partially because they are to aggressive at aiming for and achieving high margins. They tend to not make very many "friends" outside of companies that make their GPUs. Microsoft, Sony, countless tablet/phone OEMS... They used Nvidia a few times, than dropped them.For Microsoft and Sony, it surely had to do with cost for the same performance. Especially given how price sensitive the console market is. For phone/tablet OEMS... Given Nvidia's power and perfomance numbers are true, they must have priced their chips far to expensive. Given there chips were a price to other high end SoCs, they must have had bad support/power/performance/etc.
Nvidia has a history of expensive pricing, I wouldn't be shocked if this is another case. As for "AMD troll" you're welcome to make statements like that. While I try to be objective as much as possible, I have a more positive view of how AMD interacts with consumers and buisness partners than Nvidia.
However, where was AMD part of NVidia v. Samsung/Qualcomm? Perhaps you want to relate the sale of Adreno is 2009? /meh
Samus - Monday, April 6, 2015 - link
My first commend would be "hindsight is a bitch" to Microsoft and Sony going with AMD because now that Maxwell is out, a PS4/XBOX One refresh with a more efficient\equal performance part would be tremendous (such as when Microsoft revised the Xenon/Jasper XBOX 360's)My second comment is NVidia would never sue AMD (or vice versa) because they have a bundled licensing agreement. The purpose of this condensed license is to prevent lawsuits because they essentially cover entire categories of patents, not specific patents themselves. It's ironic the Chinese government fined Qualcomm into a 1 BILLION settlement because the Chinese only believe in "a-la-cart" licensing, when it is well accepted to bundle patents throughout the world to prevent exactly what the Chinese government supposedly fined Qualcomm for.
http://www.dailytech.com/China+Smacks+Qualcomm+Wit...
testbug00 - Tuesday, April 7, 2015 - link
You make it sound like either company wanted to work with Nvidia. That's the problem. Nvidia screwed Microsoft on shrinks and charging the same price for the original Xbox. Nvidia failed to deliver the performance at the price they promised Sony with the PS3.Maybe it's just me, but, if I had a one in 8-10 year design, I would want to make sure everything went right. Gamecube, Wii, Wii U, Xbox 360. All ATI/AMD, all fine.
Anyhow, what does this have to do with what I was saying 0.o?! Thanks :)
Samus - Tuesday, April 7, 2015 - link
It just sounds like Nintendo knows how to negotiate a contract. And XBOX 360 was a disaster, it was inches from being the first recalled gaming console in history; 1 die shrink, 2 revisions, 3 years later and 40 watts of power reduction helped to resolve the same issues NVidia was blamed for just a few years prior with "Bumpgate" while Microsoft took all the flak.I'm not an NVidia fanboy (although I personally find it ridiculous to not consider a Maxwell card these days) but I am NOT a Samsung fan for dozens of reasons. They don't stand behind their products, and I have zero respect for companies like that. Microsoft, Intel, AMD, Nvidia, Apple, even HP Enterprise (with the very annoying exception of out-of-warranty BIOS updates) all stand behind their products. Samsung, Lenovo, Google, Logitech, to name a few, do not stand behind their products: slow to act on resolving customer complaints, denying warranty coverage for bogus reasons, discontinue software support for products often still sold and/or in-warranty, etc.
So yes, for unrelated reasons, I am rooting for the company I know has a good track record of customer support and listening to customers, opposed to the company that I feel is complete opposite. As a (ex-) customer of Samsung I got nowhere with them, so perhaps billions in damages and legal pressure will signal a wake-up call for them to change their shit around.
nathanddrews - Tuesday, April 7, 2015 - link
You can't blame or praise AMD and NVIDIA by how their customers (Microsoft, Sony, Nintendo) use their products.testbug00 - Monday, April 6, 2015 - link
On a side note, Nvidia sued them FOR NOT LICENSING their products. "NVIDIA’s claim notes that the company has been attempting to reach a license agreement with Samsung and Qualcomm since 2012, and that today’s suit is a result of their inability to come to an agreement over the last 2 years." (http://www.anandtech.com/show/8492/nvidia-files-pa...Nvidia either offered a (multiple?) costs to license their Kepler GPU and/or patents. They couldn't come to an agreement. You don't go into these things say "how much do you think is fair to pay" unless you have a weak position. I believe Nvidia thinks it has a strong position (correct or not) and hence would demand a large amount. Probably larger than they expected to get to make the other side feel like they "won" a concession. Simple negotiating tactics. =]
Samus - Monday, April 6, 2015 - link
It is also believe (but not confirmed) that Intel (and thus Apple, who was at the center of the case) signed a GPU licensing agreement after the 1.5 BILLION dollar settlement in 2011. The fact NVidia isn't going after Intel (or Apple) lends further weight to those license rumors...testbug00 - Tuesday, April 7, 2015 - link
They signed an agreement for cross-license for ALL of Nvidia's graphics patents at the time of signing. And some of Intel's patents also to Nvidia.They didn't license the 7 patents in this case specifically. Intel still should be covered by that license.
And, as for other companies. Just because they're only sueing 2 publically doesn't mean they're going after others. If they aren't going after Imagination Tech, Apple, and any sizable company that uses MALI GPUs, I would be shocked. They might not be sueing them yet, but, certainly have or are in some form of (cross)-licensing talks.
Samus - Tuesday, April 7, 2015 - link
Considering Samsung has been in hot water dozens of times for not licensing patents, and other companies you've mentioned haven't, my guess would be the other companies, as you've suggested, are licensing the patents.And you're right, Samsung probably looked at it like this: we sold 60 million devices that infringe on this IP during the period they want us to license for (let's just say...) $15 per device sold.
That's almost $1 BILLION right there. Samsungs' legal dept probably said "let them sue us and we'll negotiate it down to less than that." Samsung pays less, legal department stays funded, nVidia gets something, everyone wins.
And you can bet this trial will end in a settlement.
CPUGPUGURU - Monday, April 6, 2015 - link
Good seeing Nvidia aggressively pursuing payment from infringing Samsung and Qualcomm, why should they not pay for use of Nvidia's market leading graphics IP when Intel does.testbug00 - Tuesday, April 7, 2015 - link
The issue isn't if they pay or not. The question is how much is Nvidia asking for, and, was it willing to accept industry standard rates? The ARM rate for a full GPU designed for you to implement is under a percentage.I find it hard to believe Nvidia would accept sub 1% royalty. For better or worse.
CPUGPUGURU - Tuesday, April 7, 2015 - link
Why would ARM be any guide to industry standard when its ARM that's infringing on Nvidia's IP.Nvidia payed for the R&D and patents and ARM takes want it wants and charges Samsung/Qualcomm want it wants this in no way sets the price for a license agreement for Nvidia's IP who owns the IP in the first place. Samsung/Qualcomm/ARM need to negotiate a agreement and its now going to be more expensive because of damages and court costs.
We'll see how this plays out, for better or worse depending on whose side you're on.
Penti - Tuesday, April 7, 2015 - link
Some of the IP companies has made GPU's with the supposedly infringing IP since before some of the patents where filed. Nvidia basically claims that everything other than their chips and Intel's is infringing. Samsung, Qualcomm, ARM, Broadcom, Imagination and others own a lot of GPU IP that is valid too.CPUGPUGURU - Tuesday, April 7, 2015 - link
"In what NVIDIA calls a “favorable ruling” from the ITC, NVIDIA has received what they consider to be favorable definitions in 6 of their 7 disputed claims."NVIDIA tried for years to come to some kind of agreement, many did such as Intel but ARM Samsung, Qualcomm and others did not and will be held accountable for infringing on NVIDIA'S Patent. NVIDIA is not disputing that, "others own a lot of GPU IP" NVIDIA is protecting their patents from those who refused to come to the table and hammer out a agreement. If/When NVIDIA wins this case others will come to table and talk but it would of been better for all and less expensive to have talked before it had to be taken to court. If/When NVIDIA wins they will be negotiating on the side of strength.
Penti - Tuesday, April 7, 2015 - link
Intel agreement stems from their 2004 chipset licensing deal with Nvidia. They never tried to block the sale of Intel products or turned to ITC in their Intel dispute. They don't mention any deal with anyone else.They didn't try for years with Samsung as they didn't actually say which patents (in full, they mentioned three to them on 7 August 2012 and two more on 15 January 2014, they guess Qualcomm knew from discussion with Samsung but didn't talk to them) they wanted to be payed for until January 2014 and didn't talk to Qualcomm at all! Don't be ridiculous just read their argument in their suit instead.
chizow - Tuesday, April 7, 2015 - link
And this is where the actual relevant patents come into play, lay them out on the table for all to see. You don't go from tile-based 2D renders to fully programmable shader architectures overnight, and you can't say "well we just copied everyone else just because".Its not disputed that Nvidia invented most, if not all of the patents that are in dispute, they have some cross-licensing with ATI/AMD in that regard but the fact the dGPU race has been a 2 horse race for the last decade long before the advent of smartphones and SoCs should tell you the likes of Imagination and ARM probably don't have a lot of relevant IP to make a case....
Penti - Wednesday, April 8, 2015 - link
Imagination made their first 3D GPU back in 95 and Qualcomm bought Adreno from AMD and made lots of contributions in the OGLES2.0 era when Nvidia was absent as well as hiring lots of people and patenting graphics stuff even before they bought the Imageon line. Both Imagination and AMD has been in the game longer than Nvidia. Nvidia didn't go after Qualcomm after they bought the Adreno technologies and as they filed the suit against them they still hadn't talked to Qualcomm according to it. If they wanted to claim that they have the sole right to produce GPU's or sell GPU IP they should have done so back then around 08. They didn't dig up the patents before 7 Aug 2012 and 15 Jan 2014 and they were irrelevant when they finally agreed with Intel in 2011. Most of them patents stuff that was already publicly disclosed or on the market, which you can't do so most claims probably could be invalidated. Samsung and Qualcomm has lots of semiconductor IP which is infringed upon by Nvidia. It's just that nobody licenses every bullshit patent and mostly they are not offered to be licensed either before suits happens. When you license IP you also get sublicensing rights to the technology in many cases and even ARM had fully programmable GPU's in the mobile space before Nvidia.chizow - Thursday, April 9, 2015 - link
@Penti, you're wrong, PowerVR was always behind Nvidia and even ATI as their last commercial dGPU card the Kyro2 lacked major functionality such as Hardware T&L. By the time programmable shaders were mainstream (invented by Nvidia btw, without dispute with GeForce 3) PowerVR wasn't even a player in discrete graphics. They managed to survive in embedded designs, then suddenly start licensing IP and enjoying a resurgence in smartphones and smartdevices. Lucky coincidence!!! Unfortunately for them there is a huge gap in what they support and what their IP is capable of, so it should be very easy for the courts to sort that out.Qualcomm was named in the suit btw, and while they did buy Adreno from AMD it is becoming more and more clear they only bought their engineering team and not any of the actual IP. The fact AMD is talking about getting back into mobile graphics and is also rumored to be licensing their IP all but confirms this.
So, this again leads to the fact Nvidia has a very strong case, after all they have laid out their patents on the table for all to see and have products unbroken in timeline from their inception until now demonstrating their leadership in the industry.
Penti - Friday, April 10, 2015 - link
T&L was known before the patent in this suit was filed, S3 had their HW T&L card on the market before Nvidia sought protection ('488) and you can't patent things that are already disclosed to the public or used by someone else, though S3's implementation was broken. It probably takes more than implementing a software pipeline in hardware to actually invent stuff that's patentable too. The scope is probably too broad.Qualcomm is sued, but they (Nvidia) admit in the filing/suit that they never ever talked to Qualcomm and that they never told Qualcomm that they were infringing or what they infringed upon. Just read the suit as said. That they weren't bothered by Intel's XScale mobile processors with integrated graphics between say 2004-2006 before the sale to Marvell (or concerned about Marvell after the sell i.e. 2006-today) is not going to help. That they weren't concerned about Samsung's ImgTec use between say 2004-2011 (they licensed the tech in 2004 and onwards) is not going to help their case. That they weren't concerned about Freescale's use of PowerVR and Vivante graphics IP between 2005-today is not going to help. Neither should that they never went after TI and others for the use of PowerVR, Vivante or Mali graphics. Neither should it help that they never talked to Qualcomm despite the fact that they had a modern mobile gpu on the market before Nvidia, despite that they knew about the 2008/09 sale of the Adreno tech they never actually informed Qualcomm until they filed the lawsuit on September 4th 2014. In the future it looks like Samsung will be a minor Qualcomm customer and corporations like Microsoft, Xiaomi, Lenovo/Motorola Mobility, Huawei, LG, Sony and others is the main Qualcomm users on the market. Many of the patents hadn't been issued when Intel and Nvidia entered into their 2004 agreement – which was about chipset licensing http://www.nvidia.com/object/IO_17070.html. They also dug up a 3dfx patent to use in the suit.
Morawka - Tuesday, April 7, 2015 - link
your licensing estimates are waaaay off. Arm charges way more than that.Arm charges licensing fee's per Core. Big.Little design's means 8 cores.
The fee per core is determined by device retail cost. on a $650 phone it's more like $15 just for the cores, and another $5 for graphics ip.
The Nvidia vs Samsung case comes down to this.
Samsung was paying Nvidia licensing money for several of their patents. They been paying them for 4 years...
Then when the S5 came out, is when Samsung said, "nah, those patents are to general, we decided we shouldn't be paying for them anymore." So they stopped paying, hence nvidia gets mad and threatened to sue. Which they made good on their threat eventually.
Now nvidia's already mad, and if they are going to go through with all the trouble of a trail, then we might as well be damn sure we go after the right person. So they start at the device maker, and work their way up.
TLDR: Samsung stop paying their licensing fee's and nvidia sued to make them pay.
HighTech4US - Tuesday, April 7, 2015 - link
Quote: Samsung was paying Nvidia licensing money for several of their patents. They been paying them for 4 years...Absolute fabrication: This NEVER HAPPENED
Penti - Tuesday, April 7, 2015 - link
Nvidia didn't bundle and sell these patents to anyone including Intel which they claim is covered by their 2011 agreement. Nvidia didn't even tell Qualcomm about the patents and only disclosed some of the patents to Samsung in 2014 shortly before they filed their suit. You can clearly see that in Nvidia's filing. They just dug up obscure stuff and didn't even ask for licensing fees until Aug 2012 but without disclosing what they thought was infringing or most of the patents used in the suit.Samus - Tuesday, April 7, 2015 - link
Seriously...when the S5 came out? Then why does this case represent failed negotiations stemming from Feb 2012 (GS3-era?)I guess in your world nVidia KNEW Samsung was going to infringe to they preemptively attempted to license? This case may cover half a decade of Samsung devices for all we know...if nVidia began approaching Samsung in 2012, it could very well involve IP used in the Galaxy Tab and the Epic 4G Touch. We'll know more when the trial starts.
name99 - Tuesday, April 7, 2015 - link
Regardless of the merits of the suit, nV is under no obligation to license its patents at all, or to license them at a price you (or Samsung) find "reasonable".IF you ask for your patents to be made part of a spec (eg WiFi, ethernet, LTE) then you have to offer fair and reasonable licensing. But not in general.
If you want to go down this path of defense, you'd have to show something like nV offered these particular patents to Khronos as a more-or-less essential part of one or more of their specs. I'm unaware of anything like that.
hung2900 - Monday, April 6, 2015 - link
Apple uses Imagination's GPU in every iPhone and iPad, and if nVIDIA wins, much more money it gains, but nVIDIA didn't sue Apple.The reason is simple: this kind of patent sueing is broken in the US. Companies like Samsung is more vulnerable in a case like this, while nVIDIA can't touch Apple in the US (remember: Obama vetoed, lol)
vladx - Monday, April 6, 2015 - link
Nvidia wouldn't sue Apple anyways since they'd lose a lot more money if Apple would pick AMD over Nvidia for the GPU part.TheJian - Monday, April 6, 2015 - link
WRONG. They'll sue as soon as they have precedent set from the first case, or use that case to get apple to settle (which if rumors are true apple is already talking to them, and surely after hearing the court reads the patents NV's way so far). Why would you go after the most influential company first? Samsung is outside usa, and Qcom is your mortal enemy at this point with nothing to gain from them (samsung has a modem, fabs, screens etc, so lots to deal over there).Imagination's PowerVR architecture is named also as something they wanted blocked. One you win this, apple can't ship a phone/tablet either, so no point in hitting them until you shutdown samsung based on the same gpus. Apple won't fight in court if they know they have a total loser, they'll wisely deal as opposed to having to possibly delay a launch that would cost them a huge portion of the market to someone else (probably samsung, once suit is done and settled they'll still be shipping). Apple is likely dealing as we speak to switch to NV gpus (there is no way imagination would win a suit vs. NV, and they make so little <60mil/yr they'd die before it was over), since a victory in samsung suit means the next case would probably have a higher chance of banning products as IMG.L stuff would already be infringing via samsung suit. A court would look and see they already are infringing here "apple" so stop shipping or deal outside court immediately. I'm sure apple would like terms NOW before NV wins and has a sledgehammer in their hands for leverage to use in negotiations.
I don't think anyone would want to be stuck with a diminishing gpu return vs. Intel's side. AMD looks like they're on their way down on gpu just like cpu (can't sell current stuff, so delay on 300's for months, no profits etc), and the apus will get squeezed to death by INTEL vs. ARM war in the race down/up for both. You couldn't go AMD ONLY for long without screwing your mac/macbook etc sales. IE, the current Mac Pro has two FirePro D700's (same as W9000's) that put out 7TF, which is pretty much exactly what NV does with TitanX with far less power used (same story with M6000 probably).
http://hothardware.com/reviews/Workstation-War-AMD...
Cuda 3x faster in Adobe Premiere than AMD w/OpenCL. They pretty much take every test here but sony vegas (who uses that instead of premiere when buying CUDA cards? Only Anandtech...LOL). Check out the conclusion page on how Cuda is massively used by apps and OpenCL in some cases is just dumped (pages on sites not updated since 2012...LOL).
Anandtech chooses sony vegas because you don't want people knowing Cuda is 3x faster ;) Considering the tests results of all the benchmarks they ran (and it was vs. Kepler/fermi), you can't screw NV without screwing yourself :) It boggles the mind when I see anandtech avoid cuda like the plague vs. AMD in anything and OpenCL. They should choose the most popular apps that use both, or pit say, Sony Vegas+AMD/OpenCL running the same test/operations vs. Adobe Premiere+NV/Cuda. There are a ton of apps you can run OpenCL plugins vs. Cuda versions (3dsmax, Premiere goes either way in mercury engine, Cinema4d, Blender, Lightwave, Maya etc). Yet Anandtech keeps running stupid synthetic junk or stuff nobody makes money with (bitcoin mining, F@H etc...LOL). I digress...
One quote from the last page of the article that pretty much sums it all up:
"If you want to use V-Ray, i-ray, or Octane Render, you effectively need an NVIDIA GPU -- V-ray might technically support OpenCL, but we couldn't get it working. The company's OpenCL page hasn't been updated since 2012 and basically tells users to switch to CUDA. Other applications, like SolidWorks and AutoCAD, tilt so strongly in Nvidia's favor they may actually offset the increased price." (note maya/creo in spec favors NV now after NV fixed drivers that caused a regression in perf). For pros, price usually isn't an issue and NV pretty much wins period.
FlushedBubblyJock - Tuesday, April 7, 2015 - link
Yep, there we go - maybe total amd fanboyism can destroy the 3300% cuda advantage, then we can all be mediocre.chizow - Tuesday, April 7, 2015 - link
Exactly, I think Nvidia went after Samsung/Qualcomm first to set precedence, but also because I think they are already working with Apple to license their GPU IP. It doesn't make sense to go after Apple if they are already in talks to license and I am sure the weight of those license fees will depend largely on the outcome or probability of a favorable outcome from this initial IP case.Kyururin - Wednesday, April 8, 2015 - link
Yay Nvidia wins, AMD close shop. :) I really hope this is the case because that way I get to see Nvidia fan boys pay $9999 for a Geforce 205, or $12000 for a 50Mhz overclocked one.Michael Bay - Wednesday, April 8, 2015 - link
You`ll be writing us letters on good old paper by that logic.chizow - Thursday, April 9, 2015 - link
Some idiot was saying this 9 years ago when Intel Conroe'd AMD...we'd be paying $10000 for our next CPU etc. etc.Oh wait, none of that ever happened.
CPUs, GPUs are luxury toys, that's it, as such there's a ceiling on the price people will spend on them limited to their discretionary income. Intel, Nvidia, Samsung, Apple, they all know this. They can only charge so much for luxury toys, if they charge too much they just start hurting themselves by reducing total sales.
name99 - Tuesday, April 7, 2015 - link
How do you know that Imagination or Apple haven't ALREADY licensed from nV (or even that they don't infringe)? Neither have a history of blatantly ignoring patents.You're constructing a narrative based on how you want the world to be, not on any actual facts.
Compare, as an example the lawsuit filed after the release of Cyclone about memory-address disambiguation. That has disappeared without a trace; presumably because either Apple licensed the patent, or there was no infringement as soon as Apple explained what they were doing. (This is not so strange. When I was at Apple, we were sued by a large patent holder who was convinced that we were infringing some patent they had on how to connected decoded MPEG video frames to an underlying time base. Their lawyers explained the patent and why they thought we had to be infringing, and I told them that's very nice, and the naive way they described of solving the problem works, but it kinda sucks for the following reasons, and so therefore I used this much more sophisticated alternative. Point is, it was all a quite reasonable technical discussion: "we think you have to be doing A to solve this problem", "you're wrong, we do B to solve the problem", "ahh, you're right, NICE! OK, sorry to have bothered you." )
testbug00 - Wednesday, April 8, 2015 - link
If the license was strong enough that Apple was licensing wouldn't Nvidia have said something? Did the license agreement with Apple state it cannot be disclosed? If so, wouldn't that imply that Apple would get something out of it? A free license? A super cheap license? It seems Nvidia would have more to gain having Apple out in the open besides Intel (who never specifically licensed there patents, making them a very weak example (imho, of course)) especially given Apple's willingness to fight over patents.That would indicate the patents aren't worth very much. Doesn't make much logical sense to me. As always, I could be wrong. It seems having an Apple license and keeping quite about it is a lose-lose no matter how it happens.
tviceman - Monday, April 6, 2015 - link
Is this as close as we'll get to an anandtech GTX 960 review?Ryan Smith - Monday, April 6, 2015 - link
It's still on the list of things to do. Unfortunately it's not at the top of the list at this second.Samus - Monday, April 6, 2015 - link
lmfaoHisDivineOrder - Tuesday, April 7, 2015 - link
I suspect this just means they'll get a better deal when all parties come to a settlement. And I bet nVidia's primary motivation in this case is to compel Samsung to rely less on Qualcomm and more on them.I also expect they want a payout from Qualcomm to grease the wheels.
frenchy_2001 - Tuesday, April 7, 2015 - link
This is a very complex case at its core, dealing with:1) who is responsible for licensing IP? Chip designer, chip manufacturer or device assembler (in this case, ARM/Qualcomm, Qualcomm/Samsung and Samsung respectively)
2) are mobile processors infringing nvidia IP? This will be settled in court, but signes point to yes, as the biggest part of the desktop graphic race (and as such research) was done by the 2 horse race of nvidia and AMD. They have a cross licensing agreement, but it seems Qualcomm did not inherit from that during their purchase of Adreno.
3) What remedy will be available? Does nvidia want licensing and payment or just ban all their competition (they would be in their right to do it if they won, as they have no obligation to license those patents).
A case that will decide the fate of mobile 3D and probably fatten up nvidia's pockets.
Penti - Wednesday, April 8, 2015 - link
If you read the suit you see that they never talked to Qualcomm and the sell of the Adreno tech would not have happened if there were any intellectual property problems. Nvidia certainly use semi tech developed by Qualcomm, Samsung and ARM, and even Imagination does have a range of graphics patents which they infringe. Going to ITC makes it all look bad for Nvidia as they claim customers can happily buy Intel or Nvidia chips instead. But Nvidia was mostly absent in the early 2007-2010 era and didn't have any products (at all) to sell when the smart phone platforms (OS's) took form. Neither did Intel in that specific era and Intel was in dispute with Nvidia over cross licensing until 2011. First 3 patents were disclosed to Samsung on 7 August 2012 and the last 2 on 15 January 2014. So they can't really claim that the supposedly infringing products could have been replaced by something else, but they do try. The patents weren't used or disclosed previously and they are not really any strong patents. They might get cross licensing deals out of it, but I couldn't imagine why anyone would pay them for that. It's not like this tech was developed in secret but that is required to patent it.creepypixie - Thursday, April 9, 2015 - link
And the fact that the price differential between Intel and AMD products has been increasing ever since is - what - good for the consumer?How are CPUs and GPUs in phones "Luxury Toys"?
How is additional licensing fees, paid to nVidia, and no doubt eventually passed on to the retail price good for the consumer?