Report: Intel May Dump Nvidia, Turn To AMD For Radeon Graphics Licensing (pcworld.com)

An anonymous reader quotes a report from PCWorld: Intel could dump Nvidia for a licensing deal with AMD as the chip giant tries to prop up its patent portfolio. Currently, Intel is under a $1.5 billion licensing agreement with Nvidia, which the two companies signed in 2011. At the time, the two companies had spent years fighting each other in courts over patent licensing, and the agreement put all that litigation to rest. Intel's Nvidia deal is set to expire on March 17, 2017, and a recent report by Bloomberg claimed that Intel is now looking to cut a deal with AMD instead.

Comments

Both Intel and AMD support FOSS.

[EzInKy]

This pairing makes much more sense then Intel and nVidia.

Re:Both Intel and AMD support FOSS.

[dbIII]

This software patent shit is WHY nvidia doesn't have open drivers.

Re:Both Intel and AMD support FOSS.

[Gojira Shipi-Taro]

Sure, if you care about FOSS support more than you do actually working with games without blowing up. I no longer do. All I care about is that the hardware that I buy is going to give me the best bang for my buck when I play games on my PC, and not blow the fuck up.

Re:Both Intel and AMD support FOSS.

[EzInKy]

I care about my hardware and software working together to do what I want, no matter how many iterations of "improvements" they go through.

Re:Both Intel and AMD support FOSS.

[Anonymous Coward]

software patents makes your hardware more expensive.

Re:Both Intel and AMD support FOSS.

[EzInKy]

Playing that game is what has destined nVidia to lose in the end. Shame they couldn't employ better soothsayers. For some of us it was obvious that openness would win in the end.

Re:Both Intel and AMD support FOSS.

[bridgmanAMD]

Can you do a quick fact check ? The AMD open drivers are BSD-licensed, and are being used on FreeBSD today.

Re:Both Intel and AMD support FOSS.

[drinkypoo]

Playing that game is what has destined nVidia to lose in the end. Shame they couldn't employ better soothsayers. For some of us it was obvious that openness would win in the end.

Their low-power android GPU line isn't encumbered in the same way. Perhaps they will be able to develop it into a high-power desktop GPU line in the future, using whatever concepts in GeForce aren't under someone else's patents.

Re:Both Intel and AMD support FOSS.

[drinkypoo]

But why would they?

Because patent licensing isn't free

One man's definite of "Progress"

[Anonymous Coward]

Is another man's definition of "Horse shit"

Intel is no longer on-board

[orledrat]

THIS is how "switchable graphics" is done. Nvidia, take note!

Life Support

[alvinrod]

AMD might have a bit of an upswing once their new Zen CPUs come out next year, but they'll need to have made some serious strides because they can't afford another Bulldozer.

My guess is that Intel is hedging and looking for a way to keep AMD around in order to avoid becoming a de facto monopoly in the x86 space, which they'd rather avoid. Give AMD enough cash to keep them upright while Intel continues to rake in big profits.

Re:Life Support

[jimbob6]

Intel already pays AMD a license for the AMD64 architecture.
in fact there are several cross licensing deals between AMD and Intel.
AMD isn't going anywhere anytime soon.

Re:Life Support

[reactor451]

This is thought as the rationale for Microsoft to invest in Apple back in the 90s

Re:Life Support

[perpenso]

This is thought as the rationale for Microsoft to invest in Apple back in the 90s

Well that worked, who is talking about the Microsoft monopoly these days. :-)

Re:Life Support

[alvinrod]

Maybe you should take a look at their financials. This is a company that hasn't had positive net income since 2011. Zen needs to be at least somewhat competitive with Intel's offerings (or they need their GPU business to take a chunk out of NV) or AMD will eventually go bankrupt.

Their stock price is so low right now that the entire company could be bought for a little over $2 billion if someone were so inclined. Intel makes more quarterly profit than AMD is worth as a company. From a certain perspective they're likely worth more if they closed shop entirely and just collected Intel's licensing fees, but Intel clearly doesn't want it to come to that.

Re:Life Support

[Mashiki]

AMD will eventually go bankrupt.

AMD has been in this position before, and Intel has bailed them out several times. Because a single company holding almost all of the market share would be bad, very bad in the eyes of regulators and anything else. One of the reasons that AMD does poorly in the eyes of investors is because unlike nvidia or intel they don't leverage their patents against other companies and generally give it away or via patent sharing. Don't forget that AMD is still recovering from the BS that Intel pulled several years ago, and Intel has yet to pay the 8B or so that the courts have ordered.

Re:Life Support

[hairyfeet]

The reports so far is Zen is 40% faster than the fastest Piledriver chip (this puts it squarely in i5, possibly i7 territory) with the first two chips being a quad with HT and an octocore with HT. If they get 16 threads for under $250? those chips are gonna sell.

And do not forget they have a complete and total monopoly on the console market, with both the PS4 and XB1 being full AMD and reports that the new Nintendo NX will be AMD that is gonna give them a HUGE advantage over Nvidia in the gaming sector as every console game is gonna be optimized for their chips.

Re:Life Support

[Anonymous Coward]

Strange enough, being effectively a monopoly makes your moves watched closely by regulators. But funding the illusion of a competitor frees Intel from the scrutiny.
In short, when you buy an Intel product, you contribute to AMD's bottom line through muddy agreements in which the amount of money paid to AMD is what is needed to keep AMD alive. With this money AMD can still produce processors, which actually aren't real competition to Intel products, but is good enough for now in the eyes of the regulators.
Maybe this was becoming a bit gross, so now Intel decides to pay AMD to have access to parts in which they are behind (GPU). However this means that at some point in the future, Intel will finally reach a point where it is ahead of AMD and Nvidia and all domains, and kill them through a combination of better performance and aggressive pricing. After this Intel will be a monopoly for both CPUs and GPUs. When the only high performance CPU and GPU designers remaining work for the same company, it will be too late.
I'm old enough to have know when the computing landscape was owned by IBM. It is sad to say that Intel has now more power on the computing landscape than IBM has ever had (at the time computers were large and expensive, bit corporations and governments were careful not to buy everything from IBM. Other contenders, like Univac, Control Data, Burroughs and others had their market share). Now buy a random PC/laptop and type lspci, on most machines, every single device except maybe one or two (the GPU when it's not integrated) is Intel's. On our most recent server, lscpi returns 164 devices, out of which 161 have Intel's id.

Re:Life Support

[Lonewolf666]

In other reports Intel currently seems to have an almost 100% advantage in IPC (Instruction per Clock) over Piledriver.
Assuming Zen gets that 40% improvement and the clock speeds are roughly the same as Skylake,
-an Intel quadcore would still win by ~40% over an AMD quadcore
-but an AMD octacore would beat an Intel quadcore by ~40%, if the application scales well to eight cores.

So it would depend a lot on the software. But AMD would finally be able to win against the i7 quads with software that scales well to eight cores.

I guess Intel could still counter by lowering prices of the E-Series. But they wouldn't like that one bit I guess ;-)

Re:Life Support

[Billly Gates]

They don't suck that bad. 40% bump will bring them within 2013/2014 haswell i5 46xxx and 47xxx territory. What Skylake has over haswell is integrated wifi, usb type c/thunderbolt3, and compared to AMD raid intel rst and insane power efficiency.

If AMD can't offer this it is dead and obsolete as OEMs are busy targetting the surface making tablets and hybrids with all these features and 10 hour battery life.

No one cares about the geek gamers rig.

Re:Life Support

[Lonewolf666]

The geek gamer market might have shrinked a bit, but there is still money in it. Besides, the requirements for the architecture are not so different from the server/workstation market. A succesful Zen processor might also work well for small servers.

Re:Life Support

[Mashiki]

APU's will never beat a GPU/CPU independent configurations, and if you're not gaming the basement priced out intel-all-in-one APU is just fine for every day work. The computer market is very slow here in the west, outside of gaming PC's. However outside of the west, especially in countries like Japan, China and various places in Central/South America or ex-bloc countries in Europe. The PC market is what it was like here back ~20 years ago, just entering it's golden age of cheaper hardware. Especially in places like Japan and China where gaming was mainly on consoles or heavily restricted.

Re:Life Support

[Billly Gates]

No one will TOUCH a non Intel cpu for a server. Too much risk and a reputation of unreliable for many PHB who remember some of the shitty VIA and nforce chipsets of athlonXP's last decade. Windows Server and Linux are well supported with Intel cpus and chipsets.

My point was a fast CPU that is about as fast as 2014 won't cut it for OEM sales if it doesn't support thunderbolt 3 aka USB type c, insane power efficiency, and SOI silicon on a chip features to cut down on size for tablet use.

The MS surface is very popular and hot now outside of what is said on slashdot and executives want small, light, and power efficient yet fast. Can AMD deliver? The Xeon for servers is a class on it's own and I do not think AMD has a chance unless they sell some desktop cpus for many years and have some whitepapers and other people testing out things before they trust it in the datacenter.

Re:Life Support

[lsatenstein]

AMD might have a bit of an upswing once their new Zen CPUs come out next year, but they'll need to have made some serious strides because they can't afford another Bulldozer.

My guess is that Intel is hedging and looking for a way to keep AMD around in order to avoid becoming a de facto monopoly in the x86 space, which they'd rather avoid. Give AMD enough cash to keep them upright while Intel continues to rake in big profits.

Intel needs a second source, in order to remain a government supplier or as a supplier for large orgs.

Re:Life Support

[sydbarrett74]

Agreed. When x86 represented the vast majority of corporate and consumer sales -- as recently as 2008-2009, before the iPhone and iPad made smartphones and tablets mass-market products -- the regulatory argument held more water. Now that ARM-family chips ship in so many devices, x86 isn't the dominant CPU architecture it once was. Intel's competition isn't AMD, but rather ARM licensees -- hence their huge push into beefing up their Atom line. As ARM64 use explodes, watch Intel's server footprint decline similarly.

Re:Life Support

[hairyfeet]

Via? Nforce? Uhhh you DO know that those haven't been made in over a decade, yes? Nobody is gonna judge a chip by chipsets that haven't been made in over a decade, if you want to argue that then they will be avoiding Intel over Netburst since that is the same time period we are talking about.

What server and workstation buyers DO care about is more threads, as a good 90% of their workloads are heavily parallel and if AMD can put out a 16 thread CPU for under $300 and if reports are correct a 64 thread workstation chip for under $500? Those are gonna sell like hotcakes for video/audio creation,data servers,render boxes, hell I could sit here all day listing roles where having that many threads (if the reports are true and we're looking at Skylake levels of performance) would be a huge selling point.

Second time this has been reported in as many days

[DivineKnight]

Stop, do not pass go, you lose 20% of your stock holdings.

Dream on: a standard GPU instruction set

[jcdr]

Probably just a dream, but this could be a very big step forward. The lack of a standard GPU instruction set have paved the way of dozen of different architectures that each consume ressources in support for a very average quality and very few open source one. A GPU architecture as standard and open as CPU would allow to concentrate the ressource on a open and high quality support.

Re:Dream on: a standard GPU instruction set

[trek00]

so you kill the innovation, like the x86 standard tied down the CPUs evolution until the born of amd64 instructions set

you need a standard graphic stack, not locked hardware: even on the same brand, the architecture completely changes quickly as more power and new features needs different technology to be accomplished

if the graphic stack continues to be supported and compatible, you got no problems (yes this is not the case of ms-windows drivers as they are never updated, like 32 bit devices on 64 bit ms-windows)

Re:Dream on: a standard GPU instruction set

[jcdr]

Someday the CPU and GPU instruction set need to merge at least in part to allow efficient architecture because there architectures difference will shrink. That don't prevent to extend the instruction set to get more performance. The important point is to not lock the instruction to only a single implementation.

AMD?

[Thor Ablestar]

Remembering a FreeBSD Radeon KMS hell I'd prefer anything else.

Re:AMD?

[laffer1]

This. NVIDIA FTW on FreeBSD.

Not sure I understand this

[rsilvergun]

do AMD and nVidia have a bunch of overlapping patents that would let Intel slot one in for the other or something? At any rate the one thing AMD does right is integrate graphics, it's just that integrated can never really compete with Discrete. My GPU is mid-range and still has it's own power supply, fans, and cranks heat out back the case...

Oh, and WTF is up with AMD's stock? It's under $3. Is there something I don't understand here? Their patent portfolio alone makes them worth more than that.

Re:Not sure I understand this

[ajarin]

You should check it first before you buy that thing

Re:Not sure I understand this

[Pulzar]

Oh, and WTF is up with AMD's stock? It's under $3. Is there something I don't understand here? Their patent portfolio alone makes them worth more than that.

They've got several billion dollars of debt, which negates much of what they have on the positive side.

Re:Not sure I understand this

[alvinrod]

It's that low because the company hasn't been profitable in years. Their patent portfolio is likely worth more on its own, but the rest of AMD that bleeds money that's attached to that patent portfolio isn't worth anything at the moment. Maybe that changes and they can be competitive again with their new CPU architecture, but they're about $2 billion in debt right now so they've got a lot of work to do in order to get back into a good place.

Re:Which is better? ATI or Nvidia?

[Anonymous Coward]

Can someone tell me which one is better, please?

Thank you !

Well now that we no longer have to consider GPU based virtual coin mining and only consider gaming, NVIDIA is better.

I know, I know, that breaks the hearts of *some* FOSS advocates but I was talking about gaming, not licensing based political agendas.

3d graphics is like VLIW

[Anonymous Coward]

In the late 90s, intel figured they could change around the instruction set for good performance, if they had good compilers. That didn't work out in reality. However, that DOES work out in 3d graphics. The problem is that the animation people want to do different things, the number of transistors keeps changing, and Microsoft changes around its graphics API and operating system.

I would like a standardized framebuffer, or something like that.

Re:3d graphics is like VLIW

[jcdr]

You are right that GPU started with architecture where the compiler take all the optimisation decision to allow the higher density of ALU into the GPU. I observe that today GPU tend to have more and more dynamic optimisation in there architecture, and I think this trend will continue. I will not be surprised that a some point in the future the GPU and the CPU will share a subset of the same instruction set. A such architecture will radially simplify the complexity of handling the compilation path for GPU compared to the actual state of CPU.

Pretty much...

[AmazingRuss]

... The definition of conservatism.

The debt can be wiped

[rsilvergun]

with a few clever legal tricks and the value of the company plundered. I'm surprised nobody is doing this. I've heard Intel keeps them around for 'competition' so they don't get a real lawsuit. Microsoft might prop them up too. But right now they look primed to be 'Bained'. I guess it depends on who they owe money too though.

Re:The debt can be wiped

[SumDog]

Keep in mind they are in every XBoxOne and PS4

Re:Which is better? ATI or Nvidia?

[Anonymous Coward]

1. amd looks better on paper
2. sometimes its products are actually higher performing
3. the first two are regularly made irrelevant by shit drivers

1. nvidia is way overpriced
2. they're anti open source
3. their drivers do work most of the time, on windows and on linux

Re:nVidia sucks balls

[Thor Ablestar]

Some time ago the Xorg ATI driver team decided that they would exclusively support KMS (Kernel Mode Switch) which obviously is NOT implemented in FreeBSD and anywhere except Linux. Basically it costed me US$1000 in unusable hardware since I falsely believed that my beloved Radeons would still be supported. The news of about 1 year ago are that the old console driver cannot support KMS but the new console driver does not support KOI-8r codepage which is required here in Russia. In other words, the hardware is still unusable. https://wiki.freebsd.org/Newco...

Almost the same problems plagued the Intel drivers (X cannot exit to text mode) at least when I tested it with FreeBSD 10.1. So I am forced either to use VESA drivers or install Geforces.

And I don't care about BSOD since it's a Windows thing and the Windows is almost nonexistent for me during 18 years.

Re:Thank god for a real tech story for a change.

[Plus1Entropy]

Yeah! If we're gonna have a flamewar, let's make it about technology! Like the old times!

Sega does what Ninten-don't!

Re:nVidia sucks balls

[loonycyborg]

Isn't KOI-8r obsoleted by utf8 for a long time already? Even consoles should have switched to utf8 already..

Re:nVidia sucks balls

[SumDog]

The ATI/AMD open source video drivers are really good; much better than the close source offerings from both AMD and nVidia.

Re:Sad to see the Republicans...

[subie]

You keep posting this crap but how about getting some testicular fortitude and post it using your real id. Or are you so pathetic and immature that you have to keep hiding behind Anonymously? Mommy and daddy would like you to take a shower, clean up your mess and please finally get a job and move into your own place. They would to like make use of their basement and computer again.

Re:nVidia sucks balls

[rl117]

Have you tried the new kms driver in FreeBSD 10.1 and 10.2? It was functional but had outstanding problems in 10.1, but by 10.2 it was working very well for me; that was with an HD6850. Unfortunately I upgraded to an R9 390, putting me firmly back into the "unsupported" category. Hopefully it will get support sometime in the future.

Re:nVidia sucks balls

[Anonymous Coward]

AMD WONT SUPPORT MAH 2 BIT OS NOBODY CARES ABOUT WHAAAA!!!

Jesus Christ. AMD has no obligation whatsoever to support you and your marginal OS. However, they have released more documentation for their chips than any other manufacturer, so if you insist on keep using FreeBSD, well get cracking. In the long run, as long as you can find someone willing to write FreeBSD drivers for your Radeons, you have a far bigger chance to keep them going rather than the nVidia stuff.

All of this however, raises the question of why you need $100 graphics cards in FreeBSD for.. it's hardly games, and office stuff should be fine with the integrated chips. Your sob-story smells funny.

Re:nVidia sucks balls

[Zontar The Mindless]

So one would think, but people in various countries persist in using their old one-nation-only charsets. Americans still use US-ASCII. Russians still use KOI-8r. Chinese still use Big5. And Slashdotters are still obliged to use EBCDIC.

Re:Sad to see the Republicans...

[ultranova]

What matters is the content of the message, not who says it.

However, it's impossible to judge the worth of that content before investing time and effort to read it. That means it's very easy to shut down anonymous discussion simply by drowning relevant messages under a flood of spam. Pseudonymity mitigates that problem, as well as facilitating more complex discussions.

Anyone that says otherwise has an agenda to kill anonymity.

No, but people who ascribe malevolent intent on anyone who disagrees with them are making every would-be tyrant's life easier by normalizing such behavior.

Every forum isn't 4chan, nor should they be.

Counter-example

[DrYak]

This software patent shit is WHY nvidia doesn't have open drivers.

No, it is not.

Software patents HAVE NOT prevented Intel to assemble an entirely separate team and/or subcontract,
so that the Linux drivers are a software stack : entirely separate from the Windows one and almost entirely made of free/libre opensource software. (Minus the mini scandal around recent firmware).

All at the same time as AMD has - in parallel of the old "fglrx" stack - has supported a parallel effort to build an opensource stack, by publishing and providing informations/documentation, and also by having some of the opensource developpers on their own payroll.
They have kept TWO parallel development: the software patent encumbered legacy aquired from FireGL *AND* a newer free/libre opensoure stack.
With so much success to the point that they are now deprecating the legacy stack and replacing it with a new hybrid stack where the foundation (DRM kernel driver on Linux) is opensource, and only the 3D library is either the opensource Mesa/Gallium3D or the closed source Radeon Crimson Driver (the new successor of Catalyst). With plans to to successively reduce the closed source surface to bare minimum (the clearly stated end goal is to have nearly everything opensource, baring a minimal optional part that gamers and workstation can switch if they value raw 3D performance more than software freedom).

Meanwhile Nvidia has done nearly nothing for the opensource world. They have barely put any effort to support Nouveau developers, except for a few limited documentation mostly to help support the embed Tegra series, which has only accidentally sometime helped support for Desktop and laptop GPUs.
No documentations for GPUs released, no material support for developpers, no opensource devs on Nvidia's payroll.

In short: software has never prevented Intel and AMD, why should this suddenly be an acceptable excuse for Nvidia ?

Re:Counter-example

[thejynxed]

Because unlike AMD or Intel, nvidia actually does use a majority of third-party licensed patents that they are not the owners of in both their hardware and software stacks, thus preventing your dream from being reality. They can't just up and say "Here, go use this stuff we don't own nor have permission to redistribute openly!", you silly goose, yet that is exactly what you're expecting them to do.

US-ASCII is UTF-8

[OrangeTide]

US-ASCII is the same as the lower 7 bits UTF-8. You can't actually tell if my post is US-ASCII or UTF-8 right now, because it is legally both.

Re:US-ASCII is UTF-8

[Zontar The Mindless]

Um.... Humour and stuff, like, y'know.

Re:US-ASCII is UTF-8

[OrangeTide]

It wasn't modded Funny, so it doesn't count.

Re:US-ASCII is UTF-8

[Zontar The Mindless]

I guess the EBCDIC reference sailed right past you and the mods, then.

Re:nVidia sucks balls

[loonycyborg]

Well, I'm Russian, and I don't know any place that still uses koi8r. Maybe in Fidonet somewhere..

Re:nVidia sucks balls

[Zontar The Mindless]

*sigh*

That poorly written headline...

[wardrich86]

Report: Intel May Dump Nvidia, Turn To AMD For Radeon Graphics Licensing

Gonna be pretty hard securing Radeon licensing from nVidia... maybe they should have considered switching way sooner.

Documentation

[DrYak]

nvidia actually does use a majority of third-party licensed patents that they are not the owners of in both their hardware and software stacks

The question isn't about their hardware. The hardware could very well be closed or open, that doesn't change the matter of opensource drivers. Nobody is asking nvidia for the VHDL of their chips. Intel's and AMD's hardware are closed too. Very few GPU have open cores in fact (only a few experimental).

They can't just up and say "Here, go use this stuff we don't own nor have permission to redistribute openly!", you silly goose, yet that is exactly what you're expecting them to do.

Again, the question is not (necessarily) for Nvidia to publish the source of their whole driver, including the parts that they don't own. That's not a requirement for having an open-source driver. Though it helps: Intel driver *are* released as opensource from the beginning. But it's not a requirement. AMD still hasn't released the code of the fglrx drivers (which may contain some parts for which AMD doesn't own all rights). But AMD has release documentation documenting their drivers. This documentation has helped writing open-source drivers (which at the beginning didn't share code with the fglrx official drivers. They were 2 different implementations of drivers for the same hardware, only 1 of which is opensoruce). AMD is also paying a few of the developers for opensource driver with their own money (again these are different devs than the closed source one. In fact, they were on diferent teams, before the AMDGPU driver got started. Hence the dramatic differences in code style and quality). the fglrx and open-source drivers only started sharing code recently, and that's an entirely new GPL part of the stack.