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).
MSI has to be the worst quality part maker on the market. I've had terrible experience with them.
If I was betting on it, I'd say ASUS would have the most profitable year.
"The difference between genius and stupidity is that genius has it's limits" - Albert Einstein
What about Via?
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
Why does a firm wishing to enter the x86 market need to buy licenses
They're probably alluding to possible patents held. Of course, NVidia has them in the graphics part and could leverage that anyway. Just another reason why patents need to be scrapped and replaced with a non-exclusive system of financial incentive, if we need one at all.
however did AMD come to own any
Ancient history. AMD got into the x86 market in the 80's when the USG required multiple sources for many components, so Intel was more or less forced to let them in if they wanted USG business. Once they were established they've worked on improvements themselves which they license to Intel, etc.
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.
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...
Nope. Foxconn easily is worse than MSI, and the worst by far is a group of power supply names behind Allied and Ultra (Deer is one of them).
A bullet may have your name on it but splash damage is addressed "To whom it may concern."
Intel and AMD has been using hardware x86-emulators running on top of specialized instruction sets since Pentium Pro and Athlon. The last native x86-chip in production was the AMD Geode, and that one is dead now.
But GPU and CPU is still very different things. Performance on CPUs is very dependent on branch, and random-memory access performance. GPU's don't have real-branches and only reads memory linearly. NVidia is going to need a completely new architecture, and can only reuse some of the algorithmic implementations (fast float-point operations, etc.)
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.
While Intel and PS4 are pretty much wild speculation - based on logic (Intel is specialist in cheap chip production, something Sony urgently needs for its PS3), the nVidia and x86 are based on hirings.
While I will not go as far as to say that nVidia is attempting to implement whole CPU, it could be that they are trying to put CPU emulator/accelerator on to GPU. Scrapping the shader language and allow to write/compile plain C/etc which can be run unmodified on both CPU and GPU is a huge step forward to allow hybrid/partial acceleration, scaling the technology from lowest-end to highest-end.
Both moves have logic behind them. Both are speculations. First was already denied. But let the soap opera run for few more episodes^W the Inquirer articles more.
All hope abandon ye who enter here.
Back in the day, many purchasers demanded that manufacturers of electronics had a secound source of components so you wouldn't get stuck with a product line you could no longer build. AMD was Intel's second source provider. This agreement went to court http://findarticles.com/p/articles/mi_m0EKF/is_n1961_v39/ai_13734404 and the result was a forced agreement that meant AMD had access to Intel intel.
There are places where the networks are not touching,and there are places where they are-Boeing's Lori Gunter
C allows for things that just don't make sense on GPUs. Arbitrary branching, pointer aliasing, etc. are poisonous for GPU performance.
GPUs excel at tasks that map N input values to one output value, with a minimum amount of unpredictable branches. If a task fits in this well, it is likely being accelerated already, via CUDA, Stream, CTM. If it doesn't fit, forcing it on the GPU is a waste of time.
What you want to look at are things like C++ DSELs, which create expression templates out of compile-time defined language specifications. This way, you can have a "shader language" that is evaluated at compile-time, either to a "real" shading language, or to plain old C++ code for the CPU.
This sig does not contain any SCO code.
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
Or are you implying that every office in the world has their own fab plant and I didn't know about it?
Yes. You didn't get yours? It should have arrived last month.
For large sets, this will be our guide even unto death, for the LORD will work for each type of data it is applied to...
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.
Actually AMD has been doing hardware x86-emulation since the K5. Cyrix started it all, with the 5x86.
This is basically the "I want it really badly so you should be forced to give it to me" argument used by people to justify entertainment piracy. You dressed it up a lot nicer, though. You should take it out to dance.
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!
Ancient history. AMD got into the x86 market in the 80's when the USG required multiple sources for many components
You know, in hindsight, keeping up this policy would have been a vastly more effective way of reigning in Microsoft than that ineffectual antitrust suit.
Actually I was speaking on a much less technical, and more human level. My point was that something that we all use and that has become a backbone of our society has essentially become public domain by nature of its own success.
That's not how patents work. The McCoy automatic oiler saved milions of dollars in labor back in the age of steam locomotives. The fact that just about every locomotive had an oiler and that railroads were the backbone of our transportation system in the 19th century in no way affected McCoy's patent on it.
If a job's not worth doing, it's not worth doing right.
PAE has driver issues, still doesn't allow individual processes to use more than 4GB (in a way anyone is actually going to bother with), and is a hideous hack reminiscent of the windowing extensions used in the days when the 640k barrier was a concern.
I think part of the problem is that graphics cards are rather superfluous "bling" accesories, especially at the high end, that really don't serve much of a purpose. I haven't come across a game I can't play on "high" or "ultra" setting with my $80 video card (attached to a Core 2 Duo 2.6Ghz box, w/ 6BG RAM) except Crysis. But still the market somehow supports $800 behemoths that aren't really useful to anyone but kids who think $100 UV activated piping makes their computer faster, and perhaps high end video people.
I think most people are sick of needing to spend half the price of their computer on video cards, where a simple console costs less, and somehow pulls the same graphics.
An $800 video card gets you very little improvement over a $100 one, over a span of five years, really. Its sort of like buying a super-computer to play Microsoft Solitaire on. You just need the extra processing power to show you can afford it, not that it actually is useful.
The really amusing thing is that most PC games are cheap ports of console games these days, but somehow people think they need 10x the consoles GPU to play. There are very few games made for the high end.
Yes, I speak only of games, but what are you actually using that big $800 GPU for?
BTW: GPUs are like CPUs, not like full mobos, all the other crap is separate.
A patriot must always be ready to defend his country against his government. -edward abbey