Slashdot Mirror
Nvidia Is Trying To Make an x86 Chip
Slatterz writes with a story from PC Authority which says that "Word has reached us that Nvidia is definitely
working on an x86 chip and the firm is heavily recruiting x86 engineers all over Silicon Valley. The history behind this can be summarised by saying they bought an x86 team, and don't have a licence to make the parts. 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."
What does that mean, "they don't have a licence to make the parts"? Are they not designing it from the ground up? Are chips typically made up of a bunch of simpler elements, designed by a third party?
Maybe they just want to run Quake 3 raytracing at 5fps. I mean who wouldnt?
Be you Admins? nay, we are but lusers!
Nvidia are going to challenge the concept of licensing an instruction set, and they know they are going to win.
That will be a great day for all the technology industry and herald a massive price crash in processor power.
how does pcauthority.com.au get away with re-posting others articles without even linking back to the original source (yes, I know that they credit theinquirer.net at the top, however it just links to all articles stolen from theinquirer.net).
The $2700 "gaming" CPU, coming soon from Nvidia. Combine that with your $800 twin video cards, and we're almost back to $5000 per computer again. The worst thing is, people actually buy these overpriced graphics cards giving them incentive to keep doing it. Well, have fun during the recession. I think MSI is going to make a lot of money.
Seven puppies were harmed during the making of this post.
At this point, I think it's ridiculous for any part of the x86 (or even AMD64) arch to be patentable. Almost every office on the planet has one --- you don't get much more public domain than that. However, assuming they really can't get patent licenses and can't get around that by some legal loophole, what does that leave? The only thing I can think of is that patents don't apply to software, and that they may be able to achieve decent performance running an x86 emulator on a modified instruction set GPU.
The day after he brought you news about Intel creating the Playstation 4 GPU discussed here comes more industry shaking news, original article here.
Wow, that's two pretty big news scoops on back to back days for Charlie with both making Slashdot's homepage at the same time!
Why does a firm wishing to enter the x86 market need to buy licenses, and if this is true, however did AMD come to own any if intel was the one who made x86 afaik?
Just wondering.
What about Via?
Some competition would be great, to keep Intel from reaching dominace over that market. I've feared for a long time, that if AMD should go bust prices would rise sharply.
"This sentence is false" - sending a computer to hell
I can think of a few reasons why nVidia might want a bunch of x86 engineers on-board, and they're not all "to design an x86 chip". nVidia have been pushing the GPGPU model for a while so having people around who know CPU architecture would be very useful, especially if they're looking at ways to emulate x86 assembler on their GPU architecture (which, for a few apps, would be an awesome feature).
The article is full of assumptions and conjecture. And it comes across as incredibly bitter toward nVidia. Did they turn the author down for a job or something?
http://twitter.com/onion2k
Shame about that, at least try and find some additional information and link to the original article. I didn't know that the INQ has become a news agency of sorts. They certainly don't have the credentials for that. And the author of this article even less.
Then again, we can discuss the idea that nVidia is apparently (no proof whatsoever of the hirings) going for x86 without having the licenses to do so. As I understood, AMD and Intel (and VIA) let each other use patents and designs for x86, so I assume this is about letting nVidia in or not on that scheme.
Personally I'm wondering why nVidia and VIA don't fuse. One just has created a neat little x86 CPU and low power parts the other has neat GPU's. And I heard that VIA is going out of the chipset business anyways.
See, I've started up the discussion for you. If you don't like it you can order up another if you don't think it's any good.
PC Authority ripped off this story, word for word, from The Inquirer. The author at The Inquirer, Charlie Demerjian, ought to sue their pants off for copyright infringement.
they can try to demand unrealistic prices, but they can't deny nvidia the license if they start to negotiate.
There are laws against anti-competitive behaviour and it covers exactly that!
The PC Authority site got slashdotted, but this sounds terribly like Charlie Demerijan's article from 2 days ago.
And while Charlie's articles are terribly fun to read, they don't quite qualify as news. Call them rants, speculation, whatever you wish, but not news. At least unless they get picked up blindly by other publications...
Perhaps they could be making GPGPU that with a translation layer for x86 instructions, like the Transmeta Crusoe did in VLIW, or maybe they are enhancing a Via Nano CPU design with on die GPU (rather like they did with the Tergra ARM11 chip). Either way this won't be a desktop CPU, and it won't be serious competition for Intel, but could be targeted at the growing netbook market.
Intel could step in and try to block them, but they have lost against Via and Transmeta in the past, and they would also put themselves in a difficult situation, since they are being watched in the US, EU and Asia for antitrust violations. This would look quite bad for them.
... sortof. NVIDIA has a 386(!) SoC from the acquisition of ULI.
I'm skeptical about a new entrant like NVIDIA gaining any traction in the x86 market, they would have better luck pushing out their ARM chips.
If you have the cash, intel doesnt need cash AMD does.
They may have the base architecture available, but not any of the fancy simd or 64-bit instruction sets.
First appearances (not necessarily patent dates):
MMX - 1997
3DNow! - 1998
SSE - 1999
SSE2 - 2001
AMD 64 - 2003
Intel 64 - 2004
SSE3 - 2004
SSE4 - 2006
Of course, most software doesn't use any of these extensions, but Intel and AMD can use this as a weapon in a possible FUD campaign.
AMD making GPUs and now NVidia making CPUs, maybe they plan on designing highly coupled CPU/GPU systems for better efficiency and also trying to shake both markets for more impact.
Intel has been doing GPUs for a long time, but they've done it like they do motherboard chipsets, with low interest on competing in the high-end.
What I would really like to know is who at Nvidia thinks this is a good idea? Do we really need another x86 supplier? Are they going to aim for the low end or the high end? If it's the high end, I thought that Nvidia contracted out their manufacturing. http://industry.bnet.com/technology/1000386/nvidia-chip-problems-might-be-warning-for-everyone/ Maybe that explains why the company has had trouble with some of its graphics chips in the last year or so. http://news.cnet.com/8301-13554_3-10020782-33.html
Personally, I'm a little tired of companies contracting out their manufacturing to Asia to cut costs, and then not owning up to manufacturing defects when they come to light. It has kept me from buying an Xbox 360, and it will keep me from ever buying an Nvidia CPU. Of course, I don't drink Kook-Aid, so I'm obviously not the potential customer here.
Make love, not reality television.
Why don't they try making graphic chips that don't fail? And drivers that don't crash?
I've torn apart quite a few 286 and 386 computers where the chip clearly says "Intel 286" or "Intel 386SX" and in small print down by the serial # for the processor, it is stamped "AMD" or "Made by AMD" or something along those lines (I can't exactly remember how the AMD part was worded)
I think we're at the point where x86 licensing is honestly kind of silly. For the sake of competition, I believe nVidia will find the right buttons to press and get at least enough breathing room to build parts.
Saying that x86 is a technology that allows Intel or AMD chips to run very powerful software is completely off-target. x86 is a vast software market, which chip makers continually convoluted their designs in order to have the ability to serve.
In other words, it's quite clear that x86 is not a technology anymore and has become more like a standard, which all companies should have some fair access to.
"There are some people that if they don't know, you can't tell them." ~ Louis Armstrong
They have to prepare in case Larrabee from Intel becomes a success.
Larrabee might or might not be a success but IF it does and IF intel has something very good there (maybe even in the steps Larabee II or Larabee III) that does work and has advantages everyone who has to start @zero will be out. There's no way Nvidia could catch up again and a look at the other competitor (ATI) must scare them: Ati has a vast ressource with a lot of experience and working products in the x86 field.
If the rumour is true, nvidia is doing the only useful thing: They start to accumulate experience before it's too late.
Should Larrabee (I, II, or V) work out, Ati can catch up via AMD in one or two years, nvidia would probably need more time and much more investment then.
The rumour is probably false but if it's true it makes perfect sense for nvidia to scout into x86-land.
Just 4 Gb of RAM, a 32-bit address, and make it as fast as you can. Forget about that 64-bit bullshit, I'm not running the Social Security database. But it must be on a single chip, or as close as it can be. Memory access times are limited by the speed of light once you get into the GHz range, a nanosecond is 300 millimeters.
To go with that, let's have some thousands of cores for number crunching. Mega cores, giga cores, you can never have enough cores for number crunching. But these cores need not have 64-bit capability, all they need to do is multiply-add operations, as quickly as possible.
The CPU industry, unfortunately, has been too long in a monopoly situation. Nvidia has done some impressive progress in getting an alternative thinking to the market, let's see what they can bring next.
Anyone know what x86 instructions don't need a license? I would be suprised if the 80486 instructions were still patented, but I could understand if SSE was still under patent. IMO: they really should be free to use, but most of the newer instruction sets aren't terribly needed for most tasks. I wouldn't mind a (more efficent) 486 equivalent with 512KB L2 cache running at 2Ghz :-). Of course, there are also patented technologies that don't affect instructions at all, which could not be implemented.
Objectively, what would they need a license for? There is nothing innovative in the x86 instruction set, and it has been cloned. Numerous times already.
They are doing something along the line of a high-end geode. Simply a LOW cost chip with power to run embedded systems, or netbooks. Skip all the external peripherals. Video, audio, USB, ethernet (x2). Add ram, including flash, and good to go.
I prefer the "u" in honour as it seems to be missing these days.
You might find that a lot of the technology that enables chips to run at speeds in the 2GHz region is patented.
Except that Intel and AMD hold vital patents to the set of technologies that are part of the x86 architeture. They have to cross license because they depend on each other, but they have no obligation to license to NVidia.
By stifling competition via the market positions, wouldn't they risk anti-trust charges?
There are no technological advantages of Intel's architecture over other options (say RISC). In fact, Intel's instruction set is inferior in many terms to other options out there.
There is only one reason for this situation - Windows. Microsoft tried to build VM framework named .NET that applications can use and be compiled to higher-level instruction set than x86. This could obviously open Windows to other arcitectures (say PowerPC). But they failed. .Net-only binaries are rarely seen, most programs include x86 binary code some way - for example as DLL libraries.
This means NVidia is forced to follow AMD, Intel and Transmeta and choose x86(64) to preserve backwards compatability with Windows OS.
If NVidia was smart enough, they could design their own cpu architecture from grounds, run Ubuntu on it for gaming-desktops, push this CPU into consoles market. This way, it would be finally possible to distribute games for both: PCs (gaming-ubuntu-pcs) and consoles, on single disk.
Why does it have to be x86 anyway? The only real reason I use x86 is because that's what I need to run Windows because that's what I need to play games. Nvidia is the king of gaming though. If they make a new platform on another architecture that can run games, I would be very compelled to switch to that, especially if it's Linux or BSD based. Windows is ok, but it's overpriced and I'd rather spend my money on hardware upgrades and new games than on software that runs in the background. It would be a more risky move to bypass Windows, but I think Nvidia could pull it off. On the other hand, Microsoft might budge on their x86-only policy to avoid competing with another operating system.
Intel building GPU's for gaming consoles, nVidia building a x86 CPU, Microsoft looking for OpenSource strategy.
I am confused. Or rather: I totally know what's going on.
Most earth movers only use about 50 horsepower or less. (Think about stuff you see in the city or on a farm.) Most small cars have much more than that. A lightweight sports car will boast 300 horsepower.
The older architecture is quite capable of moving mountains especially since there is a lot of existing software that is already available. Using the latest technology for a complete computer on a single chip only makes sense for the manufacturing processes, not the logic.
I still use my Pentium Pro machine. It is able to perform nicely at non-gaming tasks, well other than nethack, and most web surfing.
Cheap, reliable computing is more important than powerful computing for many applications.
Phil
Laugh, it's good for you!
Good, more competition. If Intel stop them through patent law then it shows there is a problem with patent law, or the set up of the x86 market. Competition and variation are the keys to progress. Look at nature. In the future I hope there will be real competition with desktop chip architectures. WINE with a x86 emulator could handle closed legacy x86 Windows apps..... Pretty much all *nix apps are portable and those that aren't can be made so as they are generally open.
Yes, but I bet those patents are held by foundries, not by CPU designers. The original desktop P6 implementation in the Pentium II was extended from a maximum of 300 MHz in the original Deschutes version all the way to 1.4 GHz in the Pentium III Tualatin just by a series of successive die shrinks from 350 nm down to 130 nm with almost no change in chip architecture. The PIII Tualatin has very few differences with the PII Deschutes except for SSE and on-die L2 cache. But if you think the cache made a huge difference, the 250 nm PII mobile or PII-based Celeron Mendocino with on-die L2 were only went up to 500 and 533 MHz, respectively.
So if you had a license to the 486 instruction set, you could likely build a 2 GHz version pretty easily if you made non-patentable macro-architectural changes to the original 80486 design, such as lengthening the pipeline from 4 stages to something like 12 stages and upping the FSB speed from 25-50 MHz to 200 MHz or so.
Just "gittin-r-done," day after day.
This is because, in the good old days, AMD got their start by being an official second source for official Intel chips. They were allowed to build, up through the 386, the exact same chip that Intel was building. Intel effectively gave them their start by being the "2nd source" because in those days, many companies would not consider using Intel chips until they were assured of a second source for the same chips. This was, of course, before Intel became nearly the only game in town.
Why don't they just make a SPARC chip? SPARC is an Open Standard. You don't have to license the instruction set!
/me ducks.
Stick Men
The referenced article wasn't objective at all. It was just spiteful-sounding chest-thumping. It actually reads like a screed from someone who applied for a job at NVidia and got rejected. I likely won't bother reading anything from that site again. This was in fact the first and only time I've ever been to that site before and have never even heard of it until now, which is an indication to me just how marginal and unimportant it is. Why it was ever referenced at Slashdot mystifies me.
Isn't the x86 architecture effectively public domain, after the Cyrix/IBM case?
Let's go with the Inq's speculation that nVidia are making an x86. I'm not sure why they'd want to get into a market with such established players, but the Inquirer tends to get some good gossip so let's assume they're right.
Do you really think they haven't worked out the licensing issues? A lot of companies have a cross licensing agreement. Given the huge number of patents developed by a typical semiconductor company, it simply makes sense to do this. nVidia is a big company. They spend a lot of money developing patentable technology. If GPU and CPUs are combined on the same chip they'll both want some of that tech.
The Inquirer is assuming that AMD and Intel will both bear a grudge against nVidia. It doesn't work like that. If nVidia can convince them that both sides can make a profit from the venture then they'll licence the technology. nVidia's past behaviour will be a factor but it isn't going to be a deal breaker.
So, a couple of posters have mentioned NEC, but nobody's mentioned, specifically, the V20 & V30, which, IIRC, were reverse-engineered, pin-out exact copies of the 8088&8086. As the term "reverse-engineered" implies, I don't think they necessarily had a license to do so, either.
Because 64 bits are an advantage only in some limited situations, and a performance killer in most cases.
When doing calculations, one of the greatest performance loss causes today is moving bytes from memory to CPU and back. With 32 bits you move twice as much data in the same time, a 100% gain for free.
Another huge performance penalty comes from power consumption and heat dissipation. To perform the same calculations, 64 bits need roughly four times as many logic gates than 32 bits. A 32 bit machine could be at least twice as fast as a 64 bit machine at the same technology level, a 6 GHz CPU, think of that.
So, if 32 bits is so good, then why everybody is going to 64 bits? Answer: marketing and monopoly. There is a rather limited need at the top level CPUs for 64 bit machines, mostly doing jobs that were formerly done by mainframes, such as handling very large databases. Intel and ATI are both competing for this top level of applications, and it would be too costly for any of them to develop specialized processors for such a limited market. The solution they found was to market "64 bits" as a real need for all users.
Unless you have need to manipulate files larger than 2 gigabytes in size, you are gaining nothing from 64 bits and losing a lot.
All the talk of designs and patents is completely irrelevant.
Intel is one of the best manufacturing companies in the world. They can knock out solid products in higher yields and lower costs than anyone in the chip business.
Between that and a boatload of cash in their pockets, you simply cannot win a processor war with intel. They'll cut costs, sell for less, and you'll bleed red until you die.
See: AMD.
But maybe they might want to sell an all-in-one board with their own low power or specialty product. I wouldnt expect it to be a barn burner or super cheap or have anything that other intel/amd cpu's have.
More like a counter to Intels Larrabee plans than anything.
I bet nVidia wants to make a x86 low-power CPU for mini laptops (like Acer Aspire One or Asus Eee).
Rather than developing its own kick-ass multicore processor, Nvidia chose the lazy and, shall I say, stupid route. Way to go, Nvidia. FYI, the x86 is last century's dying dinosaur, soon to join the buggy whip and the cloth diapers in the trash bin of obsolete technologies.
We need a solution to the parallel programming problem. We need a new programming model and a new type of parallel processor to support the new model. We need solutions, not me-toos and hand-me-downs. We don't need no more stinking x86 processors. Please.
How can a smaller company buy a bigger company?
Sorry, a newbie to corporate finance.
Intel, Nvidia, AMD, and ATI have known for some time that the worlds of x86 general purpose CPUs and proprietary special purpose GPUs are converging. This was the motivation behind AMD's 2006 purchase of ATI. Instead of purchasing Nvidia, Intel has decided to development its own roadmap for convergence: Larrabee. Nvidia is now scrambling for a piece of the game by developing its own x86 technology. Not only are they late, but they are also in a precarious legal position without the required x86 licenses.
My money's on Intel.
If Nvidia wanted to fight Intel/AMD on this, now would be the time. Under the Bush Administration anti-trust basically ground to a halt, the EU was pretty much the only one holding the line. Now you've got a Democrat president, and Eric Holder already confirmed as AG. A political shift is probably already under way at the Justice Department. Due partially to both Intel's success, and AMD's incompetence, Intel is dominating the CPU market right now. It wouldn't take much to draw anti-trust attention.
Basing on this old article:
http://www.theinquirer.net/inquirer/news/780/1013780/nvidia-declares-war-intel
But also on this one:
http://hardware.slashdot.org/article.pl?sid=08/04/11/1815221
It *might* seems NVIDIA is "lock & loading" its GPU to bump up in the market on hybrids (low cost stuff?). HAs AMD+ATI already done something?
In the future We'll add a GPUCPU just as easy as adding a graphics card! :-)
nop, nop, nop #VBLANK
I doubt it makes any difference legally, but from an ethical perspective, the McCoy automatic oiler really was an innovation, which made things better than previously. The x86 instruction set isn't, really. It's just that, of the many instruction sets, we've slowly converged on one as a de-facto standard for many applications. It happened to be the x86, more for social reasons than technical ones.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
There are a large class of operations whose performance is currently killed by the fact that they sometimes have to do operations that can't be run on the GPU. This requires the roundtrip to/from system memory. If there was even a reasonably passable x86 workalike onboard, it'd open up another class of operations to GPGPU.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
n/t
Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
Plenty of companies have entered the x86 market without licenses from Intel or AMD over the years, including Cyrix (now owned by VIA Technologies), IDT Centaur (now owned by VIA Technologies), and Transmeta (now owned by Novafora). Every single time the entrant had any litigation with Intel, it wound up with a settlement and still making x86 chips. Nvidia has enough of its own patents, enough cross-license deals, and enough money to fight in court that the same pattern will hold again.
You can use that and more. If you are going to have that much memory you should be getting a Pentium Pro (circa 1995) or later and an operating system that knows how to use it. PAE (processor address extension) has been around for that long. Just because 32 bit Vista still doesn't support it doesn't mean it isn't there - the Microsoft Server operating systems can use it.
The software is the limiting factor and not the hardware.
Let's assume nvidia passes the licensing issues; Can it afford its own processor fab? If not, who will make the processors?
A lot comments here suggest that nvidia is going to need the license to make the processors, where it could make the technology and strike a deal with any of existing licensed companies to make the procs.
Mod points are a dangerous tool. Abuse them wisely.
CPU designers are not the only group that make use of an instruction set. Assembler writers do too. I don't hear complaints that unlicensed assemblers are out there. It seems to me that unlicensed use of the x86 instruction set is pretty widespread.
Nullius in verba
Most of the commenters are focusing on how nvidia is going to get a license, but why not ask who will manufacture the processors instead?
nvidia could strike a deal with a company that already has the licenses required.
Mod points are a dangerous tool. Abuse them wisely.
Comment removed based on user account deletion
Nvidia should skip to the next gen and just write a gpu based OS. Something totally multithreaded. What intel is dreaming about. A parallel processing OS.
Nvidia could probably acquire AMD nowadays, seeing as how AMD hasn't been doing to well in the markets lately. Biggest roadblock might be getting approval for that merger since that would leave us with only one discrete GPU vendor to choose from.
> So why didn't the cross licensing agreement terminate when National Semi bought Cyrix? Or when VIA bought Cyrix from National? Your speculation flies in the face of actual events.
Whether or not it terminates would depend on the language in the contract. In other words, it will terminate if the contract says it will terminate and it will not if it does not.
So you have a point that words like "usually" are misleading (because the contract either says or does not say that), but there's probably no way for us to know unless someone here is a lawyer and has access to the relevant contracts.
But it's at least possible that things could go either way. After all, the parties to a contract can agree anything they want to. Even illegal, unenforceable and voidable contracts can be fulfilled if both parties simply choose to abide by them. Not every problem is solved in a courtroom.
Knowing how to make car analogies is not a substitute for knowing what torque is. Or gear boxes. Or large tires with good traction. Or heavy weight or the earth mover. Or.. ;)
If nVidia goes bust over this, they deserve it, coz it's a crappy company. And yes, Demerjian is a dick. I never read his stupid rants.
What the world really needs to save us from 1970s technology, is a completely new CPU design started from scratch. Same goes for computers in general coz "modern" PCs carry an incredible amount of legacy shit!