Nvidia announces x86 chip

If perhaps Charlie has ever written anything not about Nvidia? Im trying to imagine him writing a childrens book...
"Little Bobby wandered through the enchanted forest. The lovely flowers and trees gleamed with health. Then Nvidia came along and f*cked it up with their terrible fabrication process"

Well if Nvidia had a better track record of chip design, tighter controls on manufacturing quality and the capitol to fend of the legal armagedon Intel will surely unleash this might be a good offering to the netbook/ MID device segment.

Personally i very much dought nvidia has the resources to pull this off. Unless the score some huge partnerships to intergrate and use these i just don't see this being a smart move.

Intel's ability to cover the market from top to bottom with chips with atom on the low end and Core i7 and it's kin at the top were will the demand come from?

AMD moved to a fab-less to cut costs and is going to be hard pressed in the comming year to stay competative and profitable and that is with already having went through developing and marketing x86 chips for a while now.

If Nvidia came make a very low power x86 part for smart phones or even score the spot in the next wave of video game consoles they might be able to leverage this but with their "bump-gate" and stock price issues still looming it is looking grim.

I only hope they got some well thought out plan to turn themselves around and that this isn't some sort of last ditch effort at survivability.

...is AMD. Intel for all it's performance isn't affordable and with the recent rumors of their to-be-short-lived consumer (affordable) sockets later this year on top of the fact that all the gamer socket 775 boards are still horrendously overpriced AMD remains the only sane choice...plus socket AM3 is out now. A long life span and upgrade path is the most important thing for me and it seems only AMD has staff that determined what they could offer to people who only have $300-$500 to spend on a full system upgrade...and that is why they're doing respectable amounts of business even now.

Besides any one who wants to play Crysis for the sheer point of beating their chest has no point to prove: I can go outside and get it in RD (Real Definition!) I get playable FPS in all the games that I DO play without standing around gawking at over-texturized shrubs.

...maybe nVidia can learn from AMD/ATI's marketing department? Maybe Intel could too? Competition is welcomed when dealing with monopolistic companies like Intel and the way they've been muscling nVidia out of things I'd like to see nVidia get away with it just to see Intel be forced to compete instead of monopolize. I don't need 120-200FPS for my 60Hz LCD, just 40 FPS or better and minimized heat output from a system when I'm not gaming 95% of the time while not costing me an arm, leg, and a girlfriend in the in four or five years when I need to upgrade.

We get it.. you don't like Nvidia... Give it a rest and tell us something interesting that doesn't involve your opinions. I hate Nvidia just as much as I hate AMD and Intel but your ranting isn't going to change my mind about anything. Cheers

@Smithy. I'd say read between the lines, but it's kind of in the headline. The article was telling us about nvidia's announcement. It's pretty big news for them to officially announce an x86 chip.

Poor Charlie, he is so anti-NVIDIA that he just seems to be incapable of writing anything else. Even if NVIDIA were to start work on an x86 CPU that would be a good thing since AMD obviously can't compete against Intel (oh, poor Charlie can't bring himself to write an anti-AMD article). This article stinks of AMD/ATI fanboy status and not true journalism.

Told us that the were quitting the chipset business.. and they monopolized Macs with they chipsets.
Told us the trouble of making 55nm GPUs and they are out there (and one in my pc for you to know).
Told us that nVidia has no license, if they have it, I'm gonna remind you that for the rest of your life. Every article you write is gonna have that reminder, just for you to know.

dont know if charlie is really anti-NVIDIA but they are doing dirty business with there renaming and stuff, and besides if people come here 2 read his stuff why not? i know i enjoy his work

Titius et al. Every tirade you post at Charlie just makes him stronger.

You do know that, don't you?

I have looked into the same crystal ball as Charlie, he just hasn't divulged this information yet. The downfall of Apple is nigh, by 2012 all their nvdia based machinary will have melted, Charlie would really have liked to say 'told you so' but I did instead ;0)

You mean you smoked the same thing Charlie smokes? That's not good, you're gonna have the same visions soon...
I suggest you quit bad companies, for your own good ;)

Perhaps they will be using GlobalFoundries to make their chips. Could this solve their lack of patent issues? Is their any real bad blood between AMD and Nvdia, outside of just the fanboys?

I don't think that's gonna solve it for nvidia because the chips will still have nvidia label on them, so Intel will surely have the power to disallow them to sell them. Besides, Jen Hsun and his arrogance will definitely continue to bash Intel, so there's no way Intel is gonna give them the license either.

GO AMD AND INTEL!!! SHOW NVIDIA WHAT YOU'VE GOT!

"Little Bobby wandered through the enchanted forest. The lovely flowers and trees gleamed with health. Then Nvidia came along and f*cked it up with their terrible fabrication process"

LOL! This was by far the funniest comment I have read in a long while. Spot on!

Intel may not have a choice with giving NVIDIA an x86 license because Intel is already under deep investigate by the European Union and they don't need any more anti-competitive behaviour

Intel will not grant NVidia an X86 license.

I say again ...

Intel will not grant NVidia an X86 license.

Charlie ... we have this on good authority from several Intel sources.

We should challenge you to another bunny suit exercise??

Whilst Intel can (and will eventually) make a decent GPU ... and integrate it on the same die as their CPU, poor NVidia cannot do the opposite.

They can make a CPU which is X86 compatible however ... and after 15 years of wrangling over the patent infringement I still doubt it will see the light of day.

I may be anti-Intel in terms of their business marketing ethics but I certainly support their assertion to protect their intellectual IP.

If there was no such protection most of us would be using Chinese copies of everything.

Given the recent aquisition of a new Chinese CNC machine we now have here at work I am of the opinion that we should have just bought the Colchester instead - because the Chinese machine broke down on day one and hasn't worked since.

Just like most of their plastic toys.

NVidia's latest GPU's have a similar longevity ... given the issues with their G92 et al masks.

If they get on the EU anti-monopoly bandwagon and give Mrs. Neelie Kroes a box of candy and nice flowers.

I'm all for it; Intel should be forced to give of a licence to NV or else suffer the wraith of the New Iron Lady.

The days of cashcows are IMHO over; open up the market and give te consumer the tools they need to get in the IT world some heads turning again.

Gawd knows I'm soooo tired of this Intel/M$ shaft, my a$$ hurts. Sorry for the french.

Here is an OpenSource - OpenCore - x86.

http://www.opencores.org/projects.cgi/web/zet86/overview

I think NVidia is not really interested in an x86 itself. The x86 architecture is legacy hardware. Maybe NVidia, like AMD, are looking into combining a CPU with DSP and configurable computing. Then both companies would be in the position to compete with Intel on the "CPU" market.

His arrogance is plain and transparent... its not enjoyable, anymore.

http://www.bit-tech.net/news/hardware/2009/03/04/nvidia-reveals-plans-for-x86-cpu/1

Better job than Charlie.

Don't pay attention to the haters, they simply cannot take being wrong or having anyone voice a negative opinion about their beloved defective hardware of choice.

With any saucy article this sort of feedback cannot be helped, and really, it simply translates to more hits, so who cares? it's like those people who listen to radio shows they hate, just so they can complain, ratings are ratings.

If you make a claim, you're delusional. If you prove your claim, you're obviously on someone's payroll. If you say "I told you so" when everyone else said differently and were wrong, they say you're arrogant.

Some people's jobs are to write about certain topics...when there is something to report on those certain topics, you should expect them to be reported.

Popular media is asked for "Balanced articles", you know, make the bad guys seem slightly more innocent, the good guys not so pure. But really, Nvidia dropped the ball, covered it up and hoped it will blow away. Charlie reported the friggin news. As he is still doing. Are you gonna blame the meteorological services of the future for accurate predictions of storms?

Talk about shooting the messenger.

P.S.

Bloody mailmen, why don't they get a real job? *lol*

Thanks for keeping the heat on Nvidia Charlie. As someone who has been burned with an HP notebook with a faulty chip (that HP refuses to recognize the warranty on) I, for one, really appreciate his candor in writing about the once and former great green hope. His tone responds exactly to that of Nvidias arrogance. You get what you give in this world, and Charlie is shoveling it right back in their face. Just sign me: No longer covered in green goo.

Rumours of Nvidian X86 CPU are OLD, Today X86 Isn't Even Technically Usefull, With Big,Big Memory Loads. X64 Would Be Smarter. For same Monies as PhysX, Havok Purchase would of bought Whole AMD operation. Ever Since Paragraphs Where Banned In Commentos, Engineering Has Fallen into disorganization..Like Way Dunnington 2 billion Trannies actually turned into Nahalem thing. originally, theINQ stated Nahalem would be 700,000,000 transistors. So New cpu market is Exciting indeed today, yet Mike & Jens Could Retire together Just On Knowledge Gathered To this Point. Point Being UlteeMatic & ParaBrainer, So Go Figure.Drashek

So, they have the chip. Who will support them in hardware?

ok, they build their own motherboards with own chips.....you sure you want to use their drivers....

and lastly, WHO will change their 1366/AM3 mobo and graphics card JUST to get a all-nvidia pc that will most probably suck/unstable.

And yet again to all the GreenFace Fanboys: Why the hell do you shoot everyone that is against the people that LIE in your face about their products?

The TRUTH hurts doesn't it? :P

I notice that always when there's another dull article by this author, still people post interesting analysis of business and technology related to the topic. thank you people. as of the author, why he sound like teenager console gamer??! how can people be the so-called stupid term fanbois and not him for first being a one-sided rubbish blabber mouth?!!

I needed to start my day off with a good laugh, thanks :-)
LOL

If I didn't read it here then I wouldn't have heard about this today. Well done for doing your job properly.

And if Charlie didn't show us the revealing scientific information about Nvidia using shoddy bump material then it would make it harder for any of you guys to make a warranty claim. With Charlie's info you should find a nice letter easier to write if you are having problems with your laptop/graphics card.

It's exciting news, but after reading the bit-tech article it appears that the System On a Chip (SOC) for Mobile Intelligent Devices (MIDs) isn't a full blown core i7.

But still it will be a huge market, probably many times larger than core i7 market. MID's and SOC's are everywhere.

This is the sort of chip that will fit in mobile phones, cars, tv's, satellite boxes, possibly all-in-one future broadband boxes that do TV/HDTV/web from your tv, etc.

Would it be cool if you could just plug your TV into your router and use a wireless keyboard/mouse to surf net, movies etc? Is that where Sony are going with their money-in-a-telly idea?

But it's not. You can listen to the whole thing if you want, but there's no announcement. Just NV saying that it makes sense for low-cost systems to be highly integrated, and they see themselves doing this eventually based off their experience with Tegra. No mention if there's any project even in planning, and no mention of what architecture.

And he specifically says (29 minute mark): "it's not necessary today", because Atom + Ion does the job. "2 to 3 years down the track" is the when he talks about it starting to make sense. Still no mention of it being x86.

You sound pretty sure its not true. And mainly that seems to be based on the fact that its "2 to 3 years down the track". How long do you think it might take to design and build your first x86 chip?

If Nvidia confined their x86 compatability to the 486 then they would be clear on patents as the 486 was announced on April 10 1989. Patents only last a maximum of 20 years so by April 10 this year all the patents on the the 486 will have expired.

still waiting for charlie to get fired.

havent been to this website in months, cause it ruins my reading. i have no idea why i came on here today. its very amusing, other websites that post links to articles here like techreport.com, everyone at techreport just comments on how nobody likes charlie and they quit going to inquirer.net. its a shame and i am also an ati fan.

Wouldn't the whole point of using the Transmeta code morphing technology be to sidestep the need for an x86 license? Build something that is almost but not quite an x86 and work around the rest with the software?

No big story. The author overblows it.

If NVidia did an x86-based GPU - then it would be something. That might change the GPGPU market. But an x86 CPU integrated with a "normal" GPU into one chip for a low-end computer? Give me a break.

Hey Charlie,

The world is moving, there are real problems to be solved. Can you really sleep well at night after writing articles like this :) ?

God Bless you Charlie

Ok how can Nvidia beat chipzilla at his own game? The secret is: Do a 8/12 core proc @ 40nm (or less) w/ ultra low TDP (less than 25 watts) - add big caches (16 to 32MB on-die) and a super fast DDR3 mem controller - focus on games/rendering and CAD app's - add the ability to use more than one socket and use a special hardware/software to chain all the systems/procs (to make a rendering powerhouse). TELSA is good but for some reason it reminds me about Silicon Graphics. So the real Q is what socket will be used and where’s my multi core graphics card??? And I'm not talking about the expensive 295GTX. Ohh and of course the CPU must be cost effective for all to work.

Yet another bogus article that came up from under my layer of thining hair. Obviously I've needed to keep my hatred for Nvidia going hot inorder to keep my poor little brain warm.

The fact of the matter is NVidia HAS NOT announced a X86 chip but rather talked about an intention to enter the X86 market in a few years.

Can't talk anymore, need to focus on my next article, "How NVidia stole my hair!"

Its true. There's a a separate team working on it. From what I've heard in the cafeteria they have Transmeta people doing it. Why we are not concentrating on graphics I don't know. BTW Charlie if you've been in the SC cafeteria you should know that it used to be almost free food but now we pay full price.

...so you didn't fancy transcribing the relevant bit yourself Charlie?
What are you for exactly?

Look to the announcement earlier this week about the tie-up between Intel and TSMC, NVIDIA's manufacturing partner.

Dozens of sites besides the Inq were predicting that Nvidia would enter the x86 market for years now (and no, they didn't get their info from your articles). Once you actually predict something that wasn't blatantly obvious to the rest of the community as the sun is bright, you'll get that pat on the back you've been yearning for.

In fact, the articles you link to don't 'predict' anything. All I'm reading is pages of speculation with no concrete evidence. In your first article you predict those engineers could be used to design phone chips or STBs along with possibly x86 chips. Stabbing randomly in the dark until you hit something is not predicting anything.

In your second article you claim that Nvidia would have a working x86 chip (a CPU/GPU hybrid at possibly 45nm no less) in 2008....which they obviously don't have.

Of course, then theres the obvious question of whether you have any evidence that those Stexar engineers are even involved in this x86 initiative Nvidia has going. I imagine in the close to two and a half years since those articles were written those engineers were all moved to other areas so they could actually get something done.

Listen to this guy. He really works there: the RM is the resource manager of the software stack, it is the first layer between the NV hardware and the rest of the drivers. If he says Transmeta people are doing it, then Transmeta people are doing it. So the way Transmeta avoided the x86 license wars is the way Nvidia will avoid them.

Wow, so they really are going to try and do this... It will be interesting if they have the energy to fight this out if it comes to court blows with Intel.

As for the comment about x86 and x64... hmmm that did not seem accurate. x86 is the dominant CPU architecture now for at least the consumer market. If I understood correctly, when you used the x64 symbol you were refereeing to 64bit addressing for CPU registers and not directly the underlining CPU architect correct? And Intel supports the 64bit addressing now following AMD's getting that into the consumer's hands 1st...

Also, that comment about drivers from Nvidia I did not follow. If Nvidia has nothing else going for them they always have had superior driver support. I think it was reported that their software teams for drivers were as large or greater than the hardware engineers.

ATI has hit or miss with their drivers, but have improved since they became part of the AMD family. Nvidia has botched a few as well, but they still own the record on driver support and stability in addition to being quicker to market.

Cool article Charlie – keep up the friction. I don't mind the downing of Nvidia as much, although it is my preferred brand. I did have issues with being insulted as a customer of said goods though - and that did not happen this time. =P

Than Jen guy remind me of crazy Qaddafi

He can piss off you fanboys so nicely. I always enjoy the comments more than the article. Cry babies! LOL. Here we go, Mr. Jen Hsung decided to go bonkers and will face the chipzilla in the open. Every sane person knows he will be squashed like a green bug, but It will be worth watching. Making CPU is a lot diferent business than GPU. You can't just churn pipelines to it. If these CPUs ever get to the shelves, that is.

C'mon people! Its absolutely blindly obvious that the Charlie vs. Nvidia thing has become a running joke. He's playing with you, purposely writing artciles this way to get you wound up :P

Even websites linking to the Inq joke about it!

Don't get all stressy - its just good fun! Charlie FTW!

Why are so many people having a go at charlie? His articles have been some of the best the INQ have ever published. I enjoy them. Leave off. Its good to know that there are informed people out there who are critical of nvidia, instead of everyone trying to lick their ass all the time.

Transmeta, as someone noted, made an unlicensed x86 compatible device and Nvidia bought a license. Indeed, Transmeta got $250 million out of Intel for patent infringement - not the other way around. Even if Nvidia made an all-hardware device the Transmeta CMS x86 interpreter source code would be a very precise description of the x86 ISA that would save much of the time for a clean room project.

If Nvidia wanted to make a Transmeta-like cpu for portable devices it might actually fly since a contemporary device could use more cache than the Efficeon did and would be much faster. It could certainly compare quite favorably to the in-order Intel Atom even for power use. The Efficeon also had a Hypertransport interface and an on-board memory interface.

AFAIK, there's no reason that an array of such processors could not be made. Each would have to have its own instruction cache because of the 'learning' and optimization feature but the data caches could be the same or mirrors. I assume the learning and optimization feature might produce code not identical between devices because of differing execution histories. It would be a nice feature if a new 'Transmeta' cpu could save the optimized code between boots like the old DEC x86 interpreter did. Benchmarkers always have to realize that they have to run the code once or twice for the best results and that it does work to speed up what a user usually does.

More speculatively, another thing that Nvidia could do with such a project would be to ditch the ~40% of the x86 ISA that isn't used any more. If the Nvidia cpu wouldn't run DOS or Win 3.11 it wouldn't be much of a negative. That would save time, complexity, and expense IF it would run Win 7. Microsoft might also have to buy into the idea since they write the PC spec, not Intel nor anyone else. Maybe they insist that x86's still have to run Gates' original Basic interpreter.....

This article and many others simply prove he needs to be fired for viral reporting, and not even being right most of the time.

http://forum.beyond3d.com/showthread.php?t=52744

Go ahead and look younder, the are other threads from people who know the whole bumping process and clearing state Charlie is talking out his ass about that as well. People who work for Intel, AMD, TSMC and UMC. Charlie has yet again out down himself in the dumb department.

Intel has known for some time about Nvidia's x86 plans. Some of the former Stexar employees still have lunch with their old friends.

I'm sure ARM could do a reasonable x86 processor by sticking a x86-to-Arm decoder block upstream of the usual Arm CPU.
- a bit like the Arm-thumb
- all modern x86 processors do something similar internally anyway...
- that would probably get around any patent issues...

Talking of Patent issues, the 'x86-license' really refers to various Intel Patents
- anyone know which ones, and when they lapse?

I think this is why everyone loves the Inq. Posting seemingly ridicules rumors (for which they are then flamed).. being right about them (being flamed again, just 'cause) and still doing it in such a way that it's always fun to read, even if you don't agree.

WTG!

It would be funny if Intel decided to gobble up Nvidia for this as Nvidia did Uli for enabling SLI on their boards.

Though that would not really be a good thing either. Would serve them right though.

Got a Nforce2 board with a Barton inside but have never had a second thought about Nvidia since those days. Their head is a little big just like Creative.

... with the intel/TSMC porting agreement, it might be possible for nVidia to make Atom+Ion SOC's at TSMC in a couple of years. intel might be uncomfortable extending the license, but they might just do it to keep nVidia from making their own core.

my personal view:

nVidia is quickly trying to diversify in a big Hail-Mary play. It's stayed too focus on one product in these very difficult times and it's getting very costly to do business the same old way as they used to.

Just like they swallowed up 3dfx back in the 90s (and their awesome Voodoo line of gfx cards), their time has come.

AMD took a smart (albeit difficult) route in swallowing up ATI. Growing pains are bound to happen but all in all, theyve been doing rather well.

If this move doesnt work out quickly for nVidia, mark my words, they're going to be making deals with AMD to be bought out at an obscenely huge price (since now they do x86 chips AND silly 3D glasses n such).

Either way, its certainly interesting.

The question is who's license are they going to use? National Semiconductors (who I presume still have a license of their own, they only sold Cyrix's)? IBMs? Vias (from Cyrix)? http://www.theinquirer.net/inquirer/news/103/1031103/nvidia-has-balls-and-may-circumvent-x86-licences

Intel will of course say 'you can't do that' the same they said Via can't get Cyrix's license (http://www.articlearchives.com/computing-information-technology/computer/447315-1.html) but haven't yet sued Via or the way they complained about AMD going fabless http://arstechnica.com/hardware/news/2008/10/intel-to-examine-amd-split-for-x86-licensing-violations.ars but have now shut up.

If you dig deeper Charlie you'll probably find that Intel and nVidia are working together on this one.They both need what each other has technology-wise. It really is that simple.