Nvidia is trying to make an x86 chip

I'm not sure they don't have a license by extension... All of the companies (and divisions of companies) they've purchased. Maybe Jen was smart? Of course that x86 team could be employed for a far different purpose, like cluster computing.

Just think, not only will your Nvidia graphics card die a miserable death, soon, you can have your Nvidia CPU go the same way too, thats if Intel's lawyers don't kill it first, should be a close race...

Well i guess my last post wasn't to far off. They couldn't get into the next gen systems so now they are going to enter the cpu market. Heck if they can't get licences why not make the whole shabam...cpu ram motherboard os...then after then make a console...he he he

"Given that the firm burned about every bridge imaginable with the two companies who can give them licences, Nvidia has about a zero chance of getting one." You are talking about Intel and who would be the second company? Please tell me you were not including AMD, as they have no ability to give anyone the x86 license, they merely have access to it from Intel. My goodness for someone who has been reporting on the industry so long, this is a pretty sad mistake. I know this is intended as an Nvidia bashing article and burning bridges makes the story better, but no need to completely mess up the facts by tweaking it further and talking incorrectly about who can theoretically license x86 to Nvidia. Commentary... fine; rumors...fine; personal agenda...fine(?); twisting facts to feed the commentary and personal agendas...not so fine

With Jen Hsun and his arrogance at the top of the company, they can't succeed anyway. They'll probably try to make a fixed pipeline x86 processor. LOL And supposing they're even able to make a processor, they can't sell it without licenses. And even if they're able to, they certainly can't compete against AMD and t3h great Chipzilla. I just hate how they say processor isn't important. GTX295 can't render a damn frame if it's working with a pentium III. And after saying all this they try to make a processor, Huang has got no shame or self respect whatsoever. "Trying to chart a path of logic through their proclamations violates all the laws of thermodynamics, several laws of physics, probably several US laws" lol that was funny.

what about VIA? They have some kind of license. Couldn't Nvidia just make the bond that much stronger?

Nvidia might be trying to take Xp for few more rounds. Nvidia helped develope X86 Gaming platform, so in end court will decide royalty, NOT presently available. Its their THING, Baby. True Nvidia might go years behind & wheres software ?Mic. Sometimes these Torts are info shares & Nvidia is behind in Vista Brain, So maybe Design will be outcome.Then theres always iA64. Yet PURE Gaming Speil Might Be idea. STeWie Drashek

Can IBM Micro sell its x86 licence & Its fab altogether? Perhaps a JV between Nvidia & IBM could lead to some interesting stuffs. I don't know how deep are nvda pockets, but it seems IBM tries to disinvest from Semiconductor.

arrrrrrrggggghhhhhhhhhhhhhhhhhhhhhhhhhhhhha

I used to really enjoy Charlie's articles, and how he would continually wind up the intellectually-starved fanbois. In amongst the laughs, there have been some fantastically researched articles too. However, things are just getting ridiculous now. This isn't a news article at all. In either blind hatred for nVidia or, more likely, an attempt to simply wind everyone up and kick the rumour-mill into hyper-drive, Charlie has completely forgotten that the Inq's tag-line does actually say "Friction" and not "Fiction". I love the Inq style of reporting, and even with all their faults and broken spell-checkers, everyone else here at least manages to refrain from reporting material that was sourced solely from the depths of their own bowels. Give it a rest, hmm? At least try a little less fibre in your diet. [PS: Insert paragraph breaks as required.]

On what grounds do Intel and AMD control who can make x86 processors? What is the basis for the x86 license? Is it copyright, patent or other? Copyrights eventually expire. I think patents do too (though perhaps the USA has started granting perpetual patents).

If i could be bothered i'd trawl the Inq for a great many articles that CD has put up stating Nvidia "does x" and link them to other articles/facts where Nvidia did not actually "do x". Or the famous Charlie declaration of a multi GPU solution based on a G200 series being impossible (http://www.theinquirer.net/inquirer/news/019/1018019/nvidia-270-290-deep-trouble). GTX 295 anyone? I love the part where he says enthusiastically "and the GX2 part is quite dead." Beautiful. My GTX 295 is obviously a figment of my imagination. (PS - my dad has a HD4870 - I built his pc) No fanboi here, just a chuckle at Charlie.

They are making cpu's... You have no sense of shame, eh, Charlie? So lets's consider the past articles: - 55 nm problems (ehm they are out) - no GX2 anymore (out now) - no chipsets (ehm) - Cuda is a failure (look at top500.org at the 29th place, more to come...) What if they have a license? Just by chance, so to say. Who is the stupid then? (Refer at the above article for a clue)

Nvidia CEO certainly is an interesting guy. He let down the mergr proposal from AMD (he just wanted to be the CEO of AMD, pretty modest) and his last chance of survival. Intel is not really trying hard enough but eventually it will get it's grapphic core right and follow AMD by embedding it on it´s processors. This will be probably the last nail in Nvidia coffin. As I´m told only 4 companies are able to make x86 processors. Those are obviously intel, AMD, IBM and Via, due to cross-licencing agreements. The only chance for Nvidia to make x86 chips is buying one of these, so it´s via. Maybe there are talks that we are unaware of. Or Nvidia will make them anyway and try selling them with pirated DVDs... Who knows? Please Charlie, continue to write just as you are, a couple people among thousands of readers are nothing! Please, one of the complainers have a gaming dad! How old are him, 10?

It is surprising to me how a rumor mill site can be so out of touch with what is going on. Nvidia has no choice but to pursue x86 architecture, or go out of business. Multi-cores aren't going to scale well past 8, and that limit is right around the corner. AMD and Intel are both developing multi-scalar graphics/CPU hybrids that are focused on massive parallelism. If you are not developiong something to compete with these future processing packages, you will not have anything to sell. Discrete graphics only companies will not have a marketplace to sell in.

for more info on the x86 license http://contracts.corporate.findlaw.com/ ... 01.01.html go here! i think the pentium opens up in 2013, but who wants to copy a 20 year old processor? I've been hearing rumors for the past two years that nvidia was working on a cpu, i'm thinking that they may have something up there sleeve...

could the SPs be used to virtualize any cpu??? stupid thought?

@Titus: No, only Intel, AMD, IBM and via have licenses. AFAIK, they can't even manufacture x86 even if they buy via. The rights are not transferable as per the license agreement. That's why Intel is considering filing a lawsuit against AMD because AMD is going to hand over the task of manufacturing to it's spin off, the foundry company, which is a separate company now.

please fix the paragraph bug... paragraph bug bug bug bug... parrrrraa graph bug... gub gub gub gub garapraph gub baragraph pub... PARAGRAPH BUGGGGGGG inq fix the paragraph bug... bugggggggggggg buggggg BUGGGGGGG!!!!!!!! paragraph bugggggg FIX THE DAMN PARAGRAPH BUG INQUIRER!!!!!!!

Sonny boy charlie, Here is a scenario, Intel gives a license to nVidia for 5 years, stair AMD in the eye during the x86 license extention meeting, and say: "Drop your lawsuit Bitch" or else!

So long as their chip works with the OS's API's, isn't that enough? /n /p /n /p paragraph here please. Future CPU's should "appear" to the OS to be a single mega-fast cpu, using their multi-cores to appear as a super-wide super-fast single core. Using the extra cores to fast-forward feed the data out of order into the buffer/stack as it's being processed. NVidia has chipset and gpu knowledge, making a cpu after making a gpu can't be any more technically difficult. ARM can do it.

I, too, would like to know what the basis might be for “licensing” the x86 instruction set. As far as I know, you can copyright a string of bits, but you can’t copyright what the bits mean.

Intel is the inventor of x86, so they hold the patents. They get to decide who gets to use it. ... . . . .. . . . . .paragraph here.... @interested_party: No, you can't run a code which is inherently single-threaded on multi-cores. That's a logical limitation, not of hardware. And no, ARM can't do it either. . . . . . . . . . . . .(fix the parrrrraaagraaaphhh bugggg...)

Maybe these spastics might come up with something big at this harsh economic climate, when everyone seems to be cutting back and kicking people out - almost halting R&D; weekly. Intel kept wasting the billions in R&D; over the years with not much to show for it these bear-belly boffins kept eating doughnuts all day, with very little research. I don't know, but hope AMD or these dumb arses (nvidia) will produce some innovations like maybe integrating a gpu - don't have to be top of the line - into the cpu and it's only for processing like the external maths processor that was integrated back in the day. As for the licensing issue, well i'm sure intel or any company will always look carefully at any proposal as this could mean extra cash for them... might even free them from the anti-trust cases in various countries. And maybe change peoples perception of intel. Right now this is all hot air and bullshitting... we'll wait and see for the truth.

It´s Charlie Demerjian, is bullshit. He hasn´t facts, or sources, nothing of nothing in this "new". As well, Transmeta hasn´t a x86 license, so nvidia doesn´t need a license either. Transcoding is the key. But anyway this news is a fake, it´s Charlie, his word is worthless. Remember his "new" about a PS4 with a Larrabee GPU? YES, this "new" 2 days ago that this same day was denied by a Sony high executive... Fake over fake, ad infinitum.

Let's face it ppl nvidia is aiming to replace the Atom. After they have replaced the chipset indicating that it makes a great HTPC system. Once you remove the Atom cpu and put one of their's (a power high performer) nvidia then can capture the netbook market (after all every Tom, Dick and Mary is buying one for themselves and their kids) and the same time the HTPC market as you can hook up those new ION netbooks with a nvidia cpu inside to their HDTVs. Two markets at once with one platform. If that doesn't tittilate your shareholders nothing will.

What about a x86 GPU? or anything e.g. a GPU that understand x86 code? So that the newly graduated student can write VB code for graphic, instead of C/ASM? You know there's less and less people know how to do that day by day... Beside, you can optimized the bit of the code to run on the GPU for fast [graphic] process... [is the paragraph thing working yet? or is it a feature?]

Add one more x86 licensee: Harris Semiconductor, which once produced x286 CPU's.

Why not INQ block posts of goofs. But @interested_party is correct ARM can do x86 and Chinese GodSon2 can also do x86. Nvidia showed its ARM type of chip, NIVIDIA Tegra http://en.wikipedia.org/wiki/Nvidia_Tegra. ssj4Gogeta want to know Nvidia's "x86" chip models aswell,its secret dont tell anyone ok !?! they are egra APX 2500, Tegra 600, and Tegra 650 processors. ARM, GodSon2, NVIDIA-Tegra all can are doing x86 through emulation i.e. transaltion of x86 intructions in to their hardware compatible intruction set. The algorithms are fast enough. Afterall how much VIA could dethrone AMD/Intel depite having hardware x86. But chinese are building supercomputer with emulator based GodSon2. Comercial emulators just have to be bottom-feeder i mean feeding at near bottom of food chain like netbooks, thinandlight, ultraportable, energy-star and other fancy words to feel better in comparison to Extreme, HT, QuadFather etc. ------------ ssj4Gogeta is wrong. go see.....http://www.rapidmind.net/product.php "single-threaded applications that leverage the full potential of multi-core processors from AMD® and Intel®" ---------http://www.acceleware.com/?ifs=1 "Acceleware solutions enable users of single-threaded applications to access multi-core processing hardware"

Charlie is like a sour ex-girlfriend, desperately cursing without any morale bottom line. Give me a break, laughingstock.

"The history behind this is here and here.." Is this a joke? I clicked on both your sources (where it says here and here) and it just links back to 2 previous Inquirer articles written by Charlie Demerjian again. "More humorous is how they are going to try and spin the whole "CPUs don't matter at all" screed that they have been on lately. "They don't matter, and x86 really doesn't matter, but we are going to do it anyway just because it doesn't matter" or something. Sounds like you are quoting someone. Can you provide source? Any great journalism ending it with "or something." How about "or whatever" next time. Do they proofread your work?

Don't ST Micro have an x86 licence? So that makes AMD, IBM, Via (Centaur), ST Micro and possibly Harris/Intersil, although Harris, I believe, only produced a 286 clone. ST Micro are struggling with a $2.7bn gross debt and are cutting 4,500 jobs, so they would appear to be vulnerable to green goblins with pockets full of cash. I'm sure I've forgotten another x86 licensee that isn't Cyrix, too. Centaur made the IDT WinChip (ick!), but I swear there was another one. Does Rise Technology ring any bells with anyone?

SIS made some x86 embedded chips too (SoC)

To say that Nvidia is an arrogant company, is an understatement, in the same way that the universe can be described as "quite big".

Yes, so they (SiS) did, which led me to ALi who also made embedded stuff, who changed their name to ULi and had their x86 business bought by... nVidia! Hmm... It may be possible that they're aiming for the Nano/Atom space as Charlie says, thus embedded stuff should not really be ruled out. It also is not to be ruled out at this point that they already have a licence for the niche they're looking at and they've kept awfully quiet about it.

Sorry, Charlie didn't say it, Bruno did in the comments. What matters is it's plausible at this point.

I dont understand why this is particularly interesting - heck, a small amount of trolling through linkedin and you can see that they are working on a CPU. Whats next? Going to break the story that Sun M. is doing one also? Oh wait, can see that in linked in also. One doesnt have too be too closely embedded (pun intended) in the industry to know all of these CPUs are being staffed/worked on. Good time to be an x86 architect or verification engineer.

Forgot to mention: Quite a few of yall dont know much about x86 at all given the comments. For the bloke talking about Via - Via has an x86 lic since they own Centaur as well as the remains of Cyrix. The more obvious ploy here would be to use their patent portfolio in the GPU market as a bargaining chip in discussing a x-lic with intel. After all, the only reason people build large patent port. (outside of Rambus/Fairchiled/Etc) is to broker these kinds of deals. In the end, it wont matter - to verify the complexity of an x86 chip fully by a new company would be nearly impossible. This isnt like cranking out a simplistic ARM/XScale/MIPS chip. x86 and all its weirdnesses would take them years and years and years of verification to get a compat. processor into the market. Good luck for them. Hopefully they are doing the GPGPU schtick or perhaps a transmeta approach with a compiler frontend; however, given their open jobs - that isnt the case.

History does repeat itself. Nvidia is heading down the same path 3dfx did when it took over. Coincidence? 3dfx tried to standarize proprietary tools and this is exactly what Nvidia is doing. In the end it will fail. CUDA and PhysX will not be standards. Sorry but Intel has got them in a corner here with Larrabee. And ATI/AMD will do just fine as they are working with Intel not against. They used to be a smart, really good company until the arrogance and tossing of words started. The Fusion projects by Intel and ATI will be the beginning of the end. Who would need a second grpahics card for Phsics processing in the fusion cpu could handle it even better? I cant beleive you dont see this one coming...or at least nvidia just did and got caught with pants down scrambling to make an X86 cpu now. Intel does own Havok. And if the cpu gpu combo supports it with a larrabee then amd will follow suit. More cost effective to ehhh... it just makes sense that they have to scramble to make a cpu doesn't it? The death of PhysX and CUDA are just around the corner. The way its meant to be played will soon be the way it should have been made.

What's piqued my interest is Charlie's statement that "it's going to be fun to watch without a licence" when there have been or are multiple sources from which they could acquire, at least with much sophisty, access to one. I wonder does Charlie know they don't have access to a licence or is it just assumption? There's more than a few out there, and with all the acquisitions and mergers going on, who knows where they all are?

First of all - as someone who works in the x86 industry - I can assure you that none of these websites (inquirer, register, fudzilla, or itexaminer) have all that deep of contacts in the industry. Its pretty easy to see that with all the mis-information which is published on here as "fact" which turns out to be wrong. In addition, alot of the information that shows up on here is quite tardy and somewhat "known" to people who have any sorts of connection to peers at other companies in the industry. As for the x-lic, they are all different. It isnt like you get an x86 lic and you can do whatever you want. Certain things like buses arent shared in those. To me, the easiest way to get an x86 lic would be to buy whats left of Via and get the Centaur design team here in Austin. Anyway, this is just more ramblings from Charlie. I like his thoughts and wahtnot - but its certainly not awe-inspiring. Headhunters have more information than this article discusses. Whats next? going to break the story that AAPL is working on their own processor? yeesh.

...the other night, Rose didn't ask any hard questions. I basically stopped believing anything Jen Hsun said after he said "We (Nvidia) invented the GPU". Really I have no idea why he was interviewed in the first place, must have been a slow news day.

I wouldn't be surprised if this rumor is seeded ny Nvidia themselves. All that Nvidia is doing lately is to distract the public, so the stock price doesn't collapse when they announce their loses. The PCB cost cutting, the latest wave of rebadging, the ION push, it's all to show the investors there are some profits coming in the future. I doubt they'll be able to afford to buy anything anytime soon, and when the RMA's quadruple because of the removal of "unnecessary" QA and parts they will blame it on everyone but themselves. When EVGA defects, we'll know chapter 11 is near.

From what I now the 64bit extension of x86 is a AMD license. They have a cross-license agreement so Nvidia would need licenses from both.

paragraph 1 p paragraph 2

OK, thanks for that. I'm quite aware that the licenses differ and most are bound up in NDA so tightly that they may just vanish in a takeover, but for nVidia, this could be a moot point. As long as they have a license for the instruction set, they'll probably be happy. Anyway, Slashdot has hold of this now, so I'm sure we'll be seeing some really sensible, enlightening comments on there... :o) br Personally, I'd rather nV left Centaur alone. The Nano is a rather nice piece of kit, a quantum leap ahead of the C7 Esther, stomping on the Atom in most benchmarks (with the notable exception of PCMark, which seems to do silly things with the Vendor string). Being 64 bit capable is another plus, along with the excellent Padlock crypto engine. It just needs marketing properly. There again, perhaps a Centaur/Via/nVidia mix would add the missing exposure to the Nano, I don't know. The dual core 45nm upcoming part might just do the same. Of course, technological superiority doesn't always guarantee a winner or we'd all be running Alphas. The tech industry is a funny old game, isn't it?

Patents are valid for 20 years. The Intel 80386 (i386) was released in 1986, so the patent expired in 2006. NVIDIA can produce an i386 chip, and it will run your favorite Linux distro (compiled for i386). Moreover: if they are making plans for the next 5 or 6 years, they probably can count on the following release dates: - 2009: i486 (introduced in 1989) - 2013: i586 (introduced in 1993) - 2015: i686 (introduced in 1995)

The original X86 Patent has expired as far as I know. But to make an X86 chip you need numerous extensions (i.e. SSE, MMX, etc.) in order to have the chip process enough instructions per clock, otherwise the CPU would be dreadly slow. These patents are held by both AMD and Intel respectively, which have a cross-license agreement. As far as I know, IBM also has a cross-licence with both AMD and Intel. Nvidia must be stupid to try and make an X86 chip. So stupid in fact, that I can hardly believe that they are truely working on one. They will never get there hands on those patents unless Jen-Hsun Huang pays with his own golden jewels or what.

What if they are not trying to make a cpu? Intels larrabee is pentium 3 based and Tim Sweeney said himself that future GPU's will not have a fixed hardware pipeline but be fully programable software paths. Methinks they are just trying to be ready for that change.

If Nvidia are providing jobs for out of work engineers, I say good for them. I strongly suspect the license angle has crossed their mind...

Since IBM got a x86 license, can nvidia cobrand with IBM by nvidia creates the x86 chip design and have IBM make it and brand it as a IBM chip... with IBM selling it as an IBM chip but nvidia borrowing the license.

obvio is right. And patents are valid for 17 years which makes the 486 available already.

So what if nvidia tries to make an x86 chip. Isn't it good for us if there is more competition in the x86 space. Remember AMD a few years back when they release a fabulous chip and made Intel make a even better chip.. Yeah, competition is good for the consumer in the long run.. So if you are against, nvidia trying a x86 chip, then you are against competition. If Walmart wanted to try a x86 chip or Texas Instrument, good for them... Why you got to be an A whole all the time.

They could have came up with an original GPU then OPEN the architecture to LINUX or BSD kernel. NVIDIA would have the first true open architecture fully operable computer. Then they would have millions of code monkeys help them develop the best computer/mid/cell for free. FOSS is the future of all computing, GNU/LINUX is at the top of the FOSS list. Who can beat FREE! oh and did I mention more stable than any proprietary code/system. Why do you think microsuck is scared?

What about a co-processor to handle security? It could take advantage of an 'old-school' x86 design without the latest bells and whistles, have an immediate market, and major bragging rights. It just needs to be wide and fast. The cpu doesn't matter, just the gpu and security (lol).

I understand that you aren't exactly a friend of NVidia, and I am not one either, but they make good hardware, and most of all they and their employees have a right to live too, and you sound like you would love to seem them all die and rot in hell. Your articles are deeply and negatively influenced by personal attitude and feelings to an extent that is inacceptable for your position. Your wild guesses about NVidia exactly is trying to accomplish with their hiring of CPU experts w/o actually knowing too much doesn't make you appear all that bright either.

Funny how out website as picking up on this like it is legitimate. Do people even read the stuff they link to as a source. Charlie linking 2 sources back to 2 articles by Charlie again is not a credible source.

Does Nvidia and ATI/AMD have cross license between each other on key GPU technology? Then can Nvidia use their cross-licensing with ATI to somehow get AMD to license some x86 tech? Since AMD is building Fusion, there is probably some GPU tech that it borrows from Nvidia. Just throwing the thought out there.

Folks, ACCESS to a license (like through a cross-license agreement) does not mean you can the turn around and license it to someone else. As an example AMD has NO RIGHT to license out x86 (without Intel's consent). It all comes down to what is in the cross license agreements. So while Via, Sis and whoever else may have access to the x86 license for their own use, doesn't mean they can then enter into an licensing agreement with another 3rd party like Nvidia. So Nvidia has ONE, COUNT THEM, ONE option if they want access to the x86 license - they need Intel's consent (and they may also need AMD's consent on top of it, not instead of it). Charlie is misinformed (again) or worded things poorly (again).