news nvidia unveils project denver an arm compatible cpu for desktops and servers

Nvidia unveils ‘Project Denver’: an ARM-compatible CPU for desktops and servers

CES — Rumors have been swirling around the possibility of Nvidia developing its own x86-compatible microprocessor for some time now, given the moves by Intel and AMD to integrate graphics and chipset functions into their CPUs. However, from the top (CEO Jen-Hsun Huang) down, Nvidia consistently has had a single answer to any questions about such a possibility: “ARM is our CPU strategy,” a statement we’ve long taken to refer to Nvidia’s ARM-based Tegra SoC processors for phone and tablets. Today at its CES press conference, though, Nvidia finally pulled back the curtain its long-rumored CPU development project, and the news was stunning.

The firm has indeed been developing a CPU in secret, an effort known by the code-name Project Denver, and that CPU is intended to compete in the desktop PC and server segments. The words “high-performance computing” were even mentioned. But the processor in question will not be compatible with the dominant, Intel-owned x86 instruction set architecture. Instead, it will based on the ARM ISA that has, in the past several years, become almost entirely pervasive and incredibly widely used in mobile phones and tablets.

Huang described the Project Denver processor as a “fully custom” ARM-compatible CPU. Although Nvidia partnered with ARM and used ARM’s ISA, we get the sense that the processor microarchiteture itself may be one of Nvidia’s own making, with a higher end target in terms of performance (and presumably also power consumption) than existing ARM-compatible CPUs. We have almost no additional details on this CPU at present, but we can now see quite clearly the outlines of how Nvidia intends to use its chip design prowess to compete with Intel and AMD.

By virtue of its ARM compatibility, the Project Denver processor should be capable of running a range of popular operating systems out of the gate, including Google’s Android and Apple’s iOS. In addition, Huang hinted strongly that his news might be related to a possible announcement coming soon from Microsoft about a version of Windows tailored for ARM-compatible processors. Huang also pointed out that developers should be able to take advantage of the tools and development environments of the existing ARM ecosystems when developing applications for Nvidia’s new CPU.

So ARM is indeed Nvidia’s CPU strategy, but that strategy is not confined to pre-baked, licensed CPU cores aimed at low-power mobile devices. Instead, Nvidia intends to take ARM into new territories.

There’s much we don’t yet know about the Project Denver CPU, its projected performance, or when it might come to market. Nvidia’s press release does mention that the CPU core will be “fully integrated on the same chip” as an Nvidia GPU, which makes sense. At least we now know what the rumors were about, and we’re deeply intrigued to see how this story plays out from here.

0 responses to “Nvidia unveils ‘Project Denver’: an ARM-compatible CPU for desktops and servers

  1. 1.) This only matters for datacenters. File Systems are a moot point for the majority of desktop users.

    2.) Again, more smoke and mirrors with cherry pickings. *nix systems are also vulnerable to zero-day-exploits. Fixes still depend on whatever or not community is willing to fix the problem. Neither platform is immune to social engineering and PEBKC. 😉

    3.) Where is the proof for this? I thought the issue was a problem with x86 ISA. It only has become an issue with partitions exceding MBR limits. Datacenters which typically use *nix related solutions already confronted and dealt with the issues. It only has become a problem for desktop users with the advent of 3GiB HDDs.

    4.) This is *nix’s biggest strength at work, you get to chose what goes with the OS. As expected it can be trimmed down to a much more managable size. Example, DSL (Damm Small Linux).

    5.) This problem is mostly an issue with design. Some *nix solutions fail at this as well.

  2. As long as nvidia doesn’t buy ARM it’s all good, but if they did it would be all bad, nay terrible.

  3. I still don’t understand why intel tried so hard to alienate Nvidia, they could have fostered a very profitable relationship. Instead this triangulation between AMD, Nvidia and Intel has hurt all parties without generating fair and direct competition.

  4. Such a merger would represent such a consolidation of industry power I doubt it would ever be approved. I also think the culture and engineering differences between these companies is to great, and in the end it would deny consumers valueable choices. I love both companies for different GPU applications. Realistically I prefer AMD for games and Nvidia for consumer workstation alternatives, especially since their last architecture overhaul. I’d hate to see these choices removed from the market, I also believe consolidation would result in a huge drop off in GPU innovation.

  5. The other, better strategy, imho would have been for NVidia to sit down and talk to AMD about a merger so they can combine to actually fight Intel together. It’s really quite ridiculous that these two companies are still trying to bash each other’s brains in when the real enemy is the Goliath gleefully watching them from the sidelines. With NVidia’s hoard of cash and it’s GPU know-how, and AMD’s x86license and CPU designing experience, the two of them would make a powerhouse company that would really threaten Intel and completely take over the mid-range and high-end/scientific graphics card market-place. Of course, that would make far too much sense and require Jen-Hsun Huang to swallow a little pride, but for the cause of seriously confronting and defeating Intel, hell I’d do it, ego notwithstanding.

  6. I would love to see what happens here but I can safely say my adoption of such a technology would depend on how it effects my ability to play games.

  7. Nvidia building a cpu of this type.

    are they trying to make X86 irrelevant bypassing Intel altogether?

    keep wondering why they didn’t partner up with Via a company with product, decent product that’s been unable to penetrate the North American market since they got bounced.

  8. Actually, that is my stance. Poor wording on my part is likely to have caused some confusion.

  9. Ok. Here are some examples:

    1) Filesystem support – Microsoft in fact support only old NTFS (if you do not count obsolete FAT). Guess what… NTFS have no match to modern filesstems like ZFS, BTRFS, Ext4, etc…

    In addition, since MS support only one FS, you can not choose appropriate FS for different tasks and solutions.

    2) Updates – Look at for vulnerability reports. If you observe them in long time period, you will find out, that every Linux system have rank of 0% unpatched vulnerabilities. Microsoft systems usually oscillate from somewhere around 3-7% of reported, but still unpatched vulnerabilities. Which is funny. Microsoft evangelists usually say that Windows have less vulnerabilities than Linux. So why the hell they are incompetent to patch them immediately?

    3) Partition table and boot support – In 2011, Microsoft still use MS-DOS bootloader and MS-DOS partitioning scheme on BIOS machines, forcing the Windows users to buy EFI base motherboard to properly boot from GPT drives.

    4) System HW requirements – I think I do not have to explain it 🙂

    5) Software management – One of the worst disadvantage of Microsoft systems. Nobody, even Operating system on his own does not know, what have been installed, what files belong to what installation. Many libraries are distributed statically linked with each program, so you have the same library hundred times. If such library is vulnerable to some exploit, you rely on all program vendors to patch all those hundred versions of the same library, which never happen….

  10. Security argument is just smokes and mirrors. Every platform has its soft spots.

    The only area where *nix outright crushes MS and Apple’s solutions is, choice. With *nix, you can decide what can you load and bundle with your OS. The options are vast, so vast that it leaves novice users completely overwhelmed.

  11. That comparsion isn’t applicable. That was back when AMD had no desktop-level platform. They had relied on third parties for that (Via, ALI, SIS and ATI). AMD got their own desktop-level platform via ATI buyout. The buyout which sealed Nvidia’s hold on AMD portion of the chipset market.

  12. Someone is going to say “security” with no explanation of why it’s less secure. Just thought I’d do it for them. 😉

  13. I think you’re confusing virtualisation with emulation. Virtualisation partitions a CPU so lots of different OSs can run on it at the same time while emulation allows code written for a different CPU/machine architecture to run.

    So you can run a 64bit guest OS on a 32bit host only if you have a 64bit CPU to begin with.

    You can’t virtualise an x86 OS on an ARM cpu or vice versa.

    This said MS could be thinking of adding some kind of hardware x86 emulation to its own ARM design (the license they brought allows them to do that kind of thing).

    However I suspect it has more to do with MS wanting all windows software written in their own .NET and Silverlight.

    want your software to run on all windows? => use .NET / silverlight
    want .NET / silverlight? use Visual Studio
    want Visual Studio? use Windows
    want Windows…

    well you get the idea.

  14. With hardware virtualization nowadays… Give us a virtualized GPU and this argument is moot.

  15. If the application was pure .NET and MS implemented .NET in ARM, you’d probably be ok.

  16. Hell froze over years ago, then:
    [url<][/url<] Seriously, though, I do see how it isn't a realistic option since AMD bought ATI. (Add on top of that the loss of NVidia's x86 license.) II doubt ARM was the only option, but heading in the same direction as the majority of non-x86 developers makes sense.

  17. OK, just so I’m clear.. will Windows programs written today automatically run on a Windows 8 ARM system? Or developers would have to specifically add support for ARM CPUs?

  18. [quote<]"if you compare it to competition, it is still behind in many important areas"[/quote<] What would those "areas" be?

  19. This is part of Nvidia’s transistion to move away from specializing on gaming GPUs. They know the PC gaming market can no longer sustain the huge costs and risks involved making a performance GPU. Intel already denied them platform access via QPI/DMI licensing. They will never make a platform for AMD unless hell froze over. Their only realistic chance is to go with another vendor (ARM). Smartphones and kin are the current craze now. They just need to a platform that can compete against current solutions.

  20. This is part of Nvidia’s transistion to move away from specializing on gaming GPUs. They know the PC gaming market can no longer sustain the huge costs and risks involved making a performance GPU. Intel already denied them platform access via QPI/DMI licensing. They will never make a platform for AMD unless hell froze over. Their only realistic chance is to go with another vendor (ARM). Smartphones and kin are the current craze now. They just need to a platform that can compete against current solutions.

  21. So wrong.

    Intel has performance emphasis, power second. ARM has power first, performance second. For many scenarios from small devices to HPC farms, the latter is better.

    Why do you think one of MS’ big CES announcements was that they are building a version of Windows specific for ARM architecture?

    And, quite frankly, the desktop market does not crave performance. They don’t know performance if it bit them on the a**. The enthusiast market that does pay attention is a smigdeon of the overall desktop market.

    I think you just like to pundit for pundit’s sake.

  22. It’s obvious WaltC is someone who watches and reads so much political punditry that it has seeped into his psyche and affects his perception of reality.

    He’s suffering from Pathological Pundit Syndrome, or PPS.

    I recommend two seasons of Jersey Shore and stay away from those high-pundit news sources!

  23. “Smartphones are going to get more and more powerful”

    Or is it that general-use PCs are getting smaller… 🙂

  24. This is exactly the right strategy. Smartphones are going to get more and more powerful. Android and IoS are much cleaner systems than WIndows will ever be but Nvidia’s chip will provide better graphics for all. I know quite a few guys who don’t even travel with their laptops anymore, just a Droid.

  25. Then it’s a good thing that [url=<]Blue Waters[/url<], powered by 300,000 8-core POWER7 CPUs, is going to be one of the most energy-efficient data centers ever built.

  26. You’re forgetting that the supreme performance of x86 solutions is provided by big-price processes. To get to the high performance, x86 has always been riding the latest process technology, and the performance carried a price premium… so it was all worth it. Meanwhile, ARM has been running on less-than-cuttingedge processes, and doing a very good job, all things considered.

    Now, what do you think ARM architecture could do, performance/power wise, if it was implemented in Intel 32/22nm?

    It’s easy to claim performance superiority when it’s backed by process superiority. So far Intel has been running away from competition in process technology, but how long will it last? Most notably, a 22nm paper was missing from 2010 IEDM… usually Intel has a paper about the new process about 1year before the first products come out (e.g., 32nm paper in 2008). Is the tick tock slowing down? Will Ivy Bridge be…. late?

  27. Regardless of subject matter or whether I agree with the general point or not, that sort of writing is deeply unsettling to me. The structure and key words would not be out of place in the most one-sided political attack. The concepts “pundits”, “sensible people”, people who don’t think for themselves, people with an agenda. That is very creepy. Very unhealthy.

  28. [quote<]As to nothing beating x86 in performance, POWER7 would disagree strongly with you there.[/quote<] Power7 would get crushed if it dared venture out from under that huge cooling apparatus its hiding under. Expose that thing to general purpose benchmarks and performance per watt tests and see how good it looks.

  29. [quote<]Repeat after me: there will be *no* desktop/server transition to ARM. Ever.[/quote<] That's a very arrogant claim.

  30. I expect that MS will allow Windows to become a much cheaper and less visible software layer over time, and rely on their impressive and growing business software (Office and friends) for the real money, maybe along with Xbox.

  31. I expect it will be MS that comes up with some sort of emulation, along with .NET or native ARM versions of important software. I doubt nVidia will do it alone. Also the Linux crowd won’t need emulation.

  32. negative ghost rider, windows has been on arm for quite some time now………………

  33. I would not call Windows 7 as a great operating system. Yes, it is probably the best OS from Microsoft, but if you compare it to competition, it is still behind in many important areas.

    More competition is great a I appreciate Microsoft move to ARM. Although I do not use their software anmore, availability of Windows for ARM will definitely help to bring ARM components to the computer market, which is great for everybody. …Extept AMD and Intel 🙂

  34. I’ve been wondering why we don’t have arm desktops yet for awhile now.

    The big ‘win’ is going to be a unified platform on your desktop, tablet and phone based on an OS that supports live APP migration. The ability to drag an app live from your desktop to your tablet and back, data and all etc is going to be what sells ARM on the desktop. You won’t ‘sync’ your phone, your phone will be a literal extension of your desktop and vice versa.

  35. [quote<]And this is relevant, how? Lots of things beat x86 cpus in performance per watt, but nothing beats them in performance.[/quote<] Relevant if you're trying to pack more CPUs (and hence transactions) per blade server. ARM isn't a catch-all, but neither is x86. Much in the same way that some workloads are better suited to 12 and 24 core Opterons than Xeons, some will benefit from more slow cores than fewer fast ones. That might be ARM, or it mightn't, but it's a lot more interesting to see companies try to innovate and compete than to simply say "it can't be done" and leave it at that. Intel is trying to scale down Moorestown to smartphone levels, and nvidia (among others) is trying to scale ARM up. It's going to be an interesting few years. As to nothing beating x86 in performance, POWER7 would disagree strongly with you there. It's too easy to be hung up on x86 as a PC enthusiast, but x86 is not the be all and end all of computing. It never was, and it never will be.

  36. [quote<][i<]ARM can't match x86 on performance per core, but it certainly beats it in performance per watt.[/i<][/quote<] And this is relevant, how? Lots of things beat x86 cpus in performance per watt, but nothing beats them in performance. You may not have noticed, but the market is absolutely ape over performance--as in 400M x86 PCs are expected to sell this year--and every single one of them blows ARM out of the water in terms of performance. No contest. [i<]That's because "performance per watt" is *all* ARM was designed for[/i<], whereas x86 is designed for performance first and PPW second. Repeat after me: there will be *no* desktop/server transition to ARM. Ever. These are markets that crave increased performance year after year after year. The ARM market is *not* the x86 cpu market--it's an embedded cpu market. The distinction is plain and obvious.

  37. ARM can’t match x86 on performance per core, but it certainly beats it in performance per watt. In dense computing applications (rackservers, HPC), that is a real advantage.

    As I have said before, the desktop will be the last to transition to ARM, if ever, because of the huge installed legacy codebase. But the trend towards ISA agnostic cloud based services (Google Docs, MS Live, even Adobe is rumored to release a cloud based imaging solution) means the barrier to entry is lower than ever.

    The main reason MS is dominant on netbooks is because it strongarmed OEMs into loading Windows XP on them. It was caught sleeping when the first Linux netbooks rolled out, and it’s trying not to be caught again with ARM, but the computing landscape has shifted since then. If Chrome catches on, that will be a foothold for ISA independent computing.

  38. Hehe, I never said it was good value, only cheaper than a $1000-2000 windows tablet convertible :P.

  39. This is all such amusing speculation–I really enjoy how it brings the anti-x86, anti-Microsoft drones out of the woodwork…;)

    How easy it is to forget that when Apple transitioned the Mac from PPC to Intel x86 that it never even considered ARM as a viable alternative…;) Yes, ARM for embedded devices (smartphones, iPads, etc) is fine, but for Macs? Hell no…;) ARM doesn’t hold a ghost of a candle to x86 cpus in terms of processing power–indeed, ARM was never designed to do so in the first place.

    It’s absolutely remarkable how some people distort reality in their never ending quest to see x86/Microsoft die–even though in all of their programmed, conditioned angst they don’t have the first clue as to how this would benefit anybody. When you see Apple dumping Intel cpus in its Macs for ARMs–come back and talk, because only then will you have something worthwhile to say on this subject.

    How easy it is to also forget that ARM has been around for a long time already and it has never offered a challenge to AMD and Intel x86 cpus–ever. The nVidia announcement changes absolutely nothing in that regard. Try and digest the reality that ARM cpus are designed for an entirely different market than the personal computer market.

    Edit: wanted to add that despite the inflammatory title of this article–I know we like to generate hits, but really–Nothing JHH has said, either in this venue or the official nVidia press release–remotely indicates that “Project Denver” concerns “desktops and servers.” Rather JHH is concrete in describing the type of devices he envisions “Project Denver” will involve–none of them “desktops and servers.” Heh…;)

    You know, this debacle is shaping up just like Larrabee did–with a lot of Internet pundits who [i<]ought to know better[/i<] swearing up and down that Larrabee was something it never was--indeed, swearing it right up until the day Intel cancels the project--and even then, even faced with official Intel documentation that "Larrabee is dead," it took many of them quite awhile afterward before they could comprehend what had happened. Such was their self-induced delusion on the whole affair. As far as I and *lots* of other sensible people were concerned, Larrabee never could do what the pundits claimed--and, more importantly, what the pundits claimed Larrabee could/would do was *never* echoed by Intel in any capacity whatsoever. I find this sort of self-deception amazing, frankly, and surely do not understand it.

  40. Sad but quite probable: the future won’t bring us many generic/open systems, be it Denver or any other.

  41. The xbox360’s PPC chips… were they the culprit? Or was it the poor engineering and cooling? Don’t forget that the designs started at 90nm for CPU and GPU and later moved on to 65nm for CPU only, which helped make it cooler. Only now as one integrated package is it at, what, 45nm for CPU+GPU?

    The PS3’s PPC chips… bad ISA? Or is it hard to program for because it’s intended to be highly parallel? If you had a Pentium 4 with 7 cores and you had to use all of them to get max performance, wouldn’t that be just as difficult to program for?

    And as for as being difficult to program for, the game console SDK provided by Sony, Microsoft, and Nintendo also has a large impact on that as well.

    And as others have pointed out, other architectures were used. Motorola 68k was used for ages. There’s that lonely Hitachi SH-4 in the Dreamcast. The N64, Playstation, and Playstation 2 used MIPS-based cores. It’s hard to say that MIPS is an unknown ISA in the embedded world. Low-cost, low-power, and plentiful. And PPC was also rocking the low-power low-cost route, too.

    ARM’s just not that special except it’s the flavor of the day, honestly.

  42. Yeah, maybe ARM isn’t there for HPC stuff, but business servers like web clusters…

    [url<][/url<] [url<][/url<] I dunno, looks like they are getting serious about giving it a go!

  43. Trying to get my head around this chip concept;
    Having a native kernel for this hardware would mean not having to recompile
    Windows software, no?
    In that case it has a chance.
    FWIW: bought some INTC stock today.

  44. On the contrary, ARM is farther away from providing HPC level designs to satisfy business on the server market, while closer to providing enough computing power for everyday users.

    Servers and high performance ARE Intel’s stronghold. “Portable” x86 (atom) has proven a failure, inefficient architecture for low power, but now, thanks to ARM (and TI, samsung, Marvell, nvidia, etc) low power doesn’t necessary mean, low performance anymore.

    McAfee’s purchase by Intel is related to intel creating future platforms incorporating mcafee intelectual property regarding enterprise level security, which is unarguably way better than their regular consumer products.

    [url<][/url<] This is where Intel thinks the future is headed towards, end to end level security. Incorporating this directly on hardware platforms is steps ahead of any current software solution.

  45. nVidia will have to include some type of emulation and would be smart to do it. However if their goal is not trying to compete with x86 and only go after what ARM already controls, this can be a very potent CPU

  46. You’re looking at it backwards.

    It won’t be x86 software being ported to (or emulated on) ARM.

    It will be existing ARM software and developers growing their current smartphone and tablet apps into a new fledgling market of desktop ARM computers, potentially displacing the established players.

    Why do you think MS is sh**ting their pants? If they were dictating terms, they would take their own sweet time about it and probably have a half-dozen half-baked projects that get cancelled before fruition (like Courier).

    Instead, they face a very real threat of hungry newcomers fresh with success from the explosive growth of smartphone software.

    Despite years of unsuccessful attempts at tablet PC computing (Windows for tablets, UMPCs), Apple beat them overnight with a technically inferior product that had all the right ingredients to succeed – UI, price, apps. They (MS) desperately need to react to this new threat, but sadly, only know how to respond in terms of corporate paradigms that may be obsolete today (Windows, Office).

  47. Agree with FP. Jen has a long history of being a visionary. And at previous points of history he was right. But as of the last 5+ years Jen has seriously been delusional. I firmly believe that his engineers and himself fully believe that transitioning windows to arm will be easy and or that Microsoft “has to do it to survive”.

    He seems awfully stuck on the “future is…” <insert crazy topic #> and trying to convince the world of it before his company goes bankrupt.

    Were i a betting man. I can see a 65-75% shot of M$ buying nvidia in the future as Jen pigeon holes his once great company further into oblivion…

  48. We’ve seen worse, Apple moved away from Motorola 6800s to PPC then from PPC to x86.
    As others said it you can have emulation for old/non CPU bound programs and either fat binaries or a different package for the more relevant/modern/CPU intensive ones.
    Third parties will follow suit if the market is large enough.

    that being said with the accent of existing ARM OSes like iOS/Android/(and WinPhone7/chromeOS to some extend as of today) Microsoft has a some catchup to do and a serious uphill battle there.

    Still times are exciting with more Hardware and Software competition than in a long time!

  49. Could somebody fill me in as to how all the software already written for Intel’s x86 code is going to run on an NVidia ARM processor?
    If they are promising some kind of emulation layer(even with MSoft’s help) I think its a recipe for disaster.
    So you buy the new processor,then select from a tiny library of newly written code to suit a desktop NVidia CPU/GPU?
    And all of this is to compete and beat(presumably)Intel?
    Yeah good luck with that Jen.

  50. Digital wrote that – not Microsoft. DEC wrote it so that software written for intel would run on Alpha. It was not part of NT 4 for Alpha.

  51. With the Core i7 series, Intel did screw Nvidia over- in the sense that they didn’t let them make something compatable with their new bus(es).

    But here’s the thing- first with the i3/i5’s with integrated video, and now with Sandy Bridge, there is no need for Nvidia to make a chipset for a consumer level system in the first place!

    And I don’t think ARM is going to challenge x86 on the desktop front, either- but I agree that it could present a huge challenge. Desktops are going the way of the dinosaur in many spaces, or are being replaced with iMac clones, that could easily run Windows on ARM. With real CPU speed from an upscaled ARM and real GPU power from Nvidia, I could see these devices easily taking over everything from the basic office desktop on down to smartphones.

  52. If you’re talking about today’s Microsoft Windows, I’m highly inclined to agree with you. But tomorrow’s? Who’s to say?

    If Microsoft is serious about bringing Windows 8 to ARM, given the breadth of their experience, what’s to stop them from making something successful with it?

    Windows 7 is a great operating system, and Windows Phone 7 is at least very effective on mobile devices (say what you will about it’s comparison to Android and iOS, and I’m an Android man myself). I really think that they have a chance here, and regardless of the final product, more competition to build better phone, tablet, and other small and portable computing operating systems is always a good thing.

  53. I assume you’re talking about smartphone/tablet gaming, or consoles? In either case, none are as powerful as what can be had in even two year old x86 CPU designs, and let’s not talk about graphics capability available on x86 platforms only.

  54. This, the 7th, generation of home consoles is the first time only PPC has been used.

    Dreamcast (gen6) used SH4. XBox was x86. In handhelds, PSP uses MIPS while DS uses ARM (as did GBA.)
    I’d say that Sony and Nintendo have been rather architecture independent in consoles.

    Microsoft had certainly used x86 to leverage developers’ experience in Windows to the XBox, but even they jumped to PPC this generation. With their recent announcement of Windows on ARM, they may be contemplating another change in processor type for the XBox’s entry to the eighth generation.

  55. I see where you’re going with this- most server applications don’t need GPUs at all, and those that do, only for monitoring.

    I think that the only instances where you might be wrong (or just short-sighted) would apply to servers that do any kind of data-intense processing, not just database-type lookups. Just as AMD is using GPU-like hardware in their CPUs to accelerate everyday processing needs, Nvidia could easily do the same here, and with an arguably better GPU platform, from a GPGPU stance.

    Also, who’s to say that Nvidia couldn’t build an SoC that focuses on CPU performance (with 8 or 16 cores, or more?), and just include a stripped down GPU core, for things like webserver and database applications? It would seem to me that they could make this technology do pretty much anything, and they have the know how to build entire systems of any type around it!

  56. I think you’re wrong about the growth, so I’ll leave it to you (or someone with better market research skills than myself) to prove your point. PC sales are increasing, and laptop sales are increasing as a percentage of PC sales. Tablet sales, with the iPad, are also increasing, but laptops and tablets aren’t mutually exclusive- many people have both, and most people with neither would choose the laptop first, because it can actually do work as well as access media and internet.

  57. I think this would be cool too. We’ll have to wait and see what happens when you open the power and cooling floodgates to the ARM community, but the idea of extensions for x86 emulation is something I like. Very, very few programs require the full power of today’s x86 processors, so if ARM can get anywhere close, working with MS they could come up with a emulation/virtualization solution that could work for almost anyone. If the app is performance limited, it’ll be recompiled for ARM anyway, so I don’t see a downside to this.

  58. Look at the history though; the Xbox was Intel x86, and the PS2 was proprietary. I think you’re right in that IBM was both reliable and inexpensive; of course, the in-order cut down version that wound up in both consoles has an incredibly low IPC, and I’m sure both manufacturers probably don’t want to repeat that.

    ARM, specifically the out-of-order version Nvidia is using for both Tegra and hopefully Denver, has the promise of being inexpensive, efficient, and fast all at the same time. Sounds like a great idea to me.

  59. Microsoft is far more dependent on Intel, than Intel on Microsoft. Don´t forget about that.

  60. “The desktop will probably remain as the last bastion of x86 for a while, but for power-limited applications (smartphones, HPC, rackservers) a switch to ARM will be a much less traumatic experience, since they’re either running Linux, already on ARM and/or run specialized code with little legacy constraints.”

    Yup… and there may be a tipping point where the floor falls out from underneath x86. Once Apple expunges the last remnants of Carbon, it will be trivial for every Mac app to be a fat binary supporting both x86 and ARM. MS is many years behind Apple in that regard, but the day that MS catches up is the day the floor falls out from underneath Intel.

  61. The only reason Microsoft Windows still hold their ground against Linux and similar OSes are Windows applications (x86) and Windows will definitely loose this “advantage” switching to ARM. In fact, there is no ARM Windows apps today in contrast to thousands and thousands of Linux apps available for more many CPU architectures.

  62. In the past, the only options for the gaming consoles were x86 and PPC. Intel x86 was too expensive and AMD x86 had an unreliable supplier. So IBM won by being a somewhat more reliable supplier than AMD and charging much lower prices than Intel. But it had nothing to do with PPC per se. In fact, let’s look at how those PPC chips have performed. The 65nm chips in the first xbox 360s ran too hot, contributing to mass hardware failures. The PPC chips in the PS3 are notoriously hard to program and to this day we have yet to see developers able to really take advantage of those chips. So there is nothing about PPC as an architecture that makes it more appealing than anything else. PPC got into game consoles based on a business decision, not a technology decision.

    Now consider ARM both from a business and technology perspective. ARM is a straightforward, low-power architecture that is well understood by a lot of developers. That sentence alone makes it superior to PPC technologically. From a business perspective, there are more producers of ARM chips than any other architecture, by far. That means a low-cost, reliable supply of chips.

    Frankly, I think it’s a no-brainer that ARM will be in at least one of the next generation consoles, maybe all of them.

  63. I would be very interested in seeing what ARM can do when there is a far higher TDP limit available to it.

    As for x86 – I’m sure that Microsoft will update their development tools to build ARM versions of software, so most modern software will be available for ARM Windows in fairly short order, including the important ones like Office. I hope Microsoft go the Apple route and have fat binaries with both code bases within, rather than requiring separate x86 and ARM distributions.

    And older x86 software would probably perform adequately via emulation – especially if NVIDIA added some features to the ISA to accelerate hard-to-emulate x86 operations, in the same way that the Chinese MIPS Longsoon processor has added them (apparently getting 70%+ of native MIPS performance for emulated x86 code, but we’ll see).

  64. Yes, but it’s MS’ boat to miss.

    With or without MS, ARM has an upswell of momentum in its favour.

    The desktop will probably remain as the last bastion of x86 for a while, but for power-limited applications (smartphones, HPC, rackservers) a switch to ARM will be a much less traumatic experience, since they’re either running Linux, already on ARM and/or run specialized code with little legacy constraints.

    Intel may make a lot of money on the desktop, but a loss of server sales and lack of forecast growth into the embedded space (eg Moorestown) would be a nasty blow.

  65. Unfortunately, it does depend on what everyone as a whole cares about.


  66. It would be ironic if intel pushing nvidia out of the x86 chipset business came back to bite them with this.

    First Microsoft, then nvidia. The smartphone space is almost entirely dominated by ARM, and it’s a major player in the tablet space as well. With almost zero growth in desktops and laptops these days, intel has good reason to be scared.

  67. [quote<] This will be great if Microsoft finds a way to make third party applications work on both ARM and x86 architectures. If not, then nVidia's chip can be all the high performance it wants to be, but it won't challenge PC's in any way I particularly care about. [/quote<] Fortunately it does not depends on what you care about.

  68. The idea of Microsoft Windows capturing ARM platform is rather funny sci-fi. The only reason Microsoft did not lost its share on x86 platform market is backward compatibility, which is exactly what the will definitely loose in case of switching to ARM.

    ARM world is dominated by Linux and even Microsoft can not change that. Linux works on ARM for decades, most of the Linux ecosystem works on ARM. I see really funny somebody thinks Windows for ARM (which do no exist today) with 0 hardware support and 0 existing applications can noticeably succeed.

  69. “The firm has indeed been developing a CPU … intended to compete in … server segments. …the CPU core will be ‘fully integrated on the same chip’ as an Nvidia GPU, which makes sense.”

    It does? Hmm, surely only fairly specialised server roles (read: small market) would require a bunch of processor space/power devoted to GPU work?

  70. Hurrah! This could be a really great move by nVidia. If any platform is going to challenge the messy ugly x86 ISA hegemony it’s got to be ARM- POWER and SPARC have lost too much steam in the past ten years. Looking forward to performance numbers.

  71. With news like this, Intel’s acquisition of McAfee is looking dumber and dumber. Time for Intel to get back to the basics and defend their territory.

  72. I’ve got to say it would be kind of interesting to see what a 65-95W+ TDP ARM processor could do…

  73. [quote=”SNM”<]Are they trying to get enthusiasts to use it instead of an x86 Windows machine?[/quote<] I doubt it. Intel and AMD have had decades and decades to get x86 and its implementations to the level of performance they currently enjoy. No way can Nvidia match that any time soon. Tegra 2 was announced to be the reference platform for future Android tablets. So taking their current relationship with Google further, it would make sense for this to be part of the reference platform for future ChromeOS smartbooks/netbooks/<whatever>books

  74. PowerPC is the architecture of choice for all the major gaming consoles. Though given Nvidia’s graphics know-how, I imagine they’d be in the best position to convince them to jump to ARM. I actually would be pretty interested in seeing Nvidia try and build a console of their own, possibly using this. I (think) they have the cash, the know-how, and the links to the gaming devs to make it happen.

  75. This will be great if Microsoft finds a way to make third party applications work on both ARM and x86 architectures. If not, then nVidia’s chip can be all the high performance it wants to be, but it won’t challenge PC’s in any way I particularly care about.

    I do think their ARM-focus could help them build a CPU for a gaming console or two. Portable or not. Netbooks will go away and be replaced by ULV laptops (Bobcat, Atom) and limited-function tablets (iPad, Android). They’ve already got tablets covered by Tegra. ULV laptops are better done by the x86 alternatives.

    nVidia is putting a lot of pressure on Microsoft. Will they REALLY support ARM rather than just claim they will and then shrug when most of the applications remain on the x86 side while other OS’s built around more limited apps do a better job of supporting ARM?

    There has to be a reason for developers to switch to ARM over x86 and how likely is it that nVidia’s ARM CPU has anywhere near the performance to force the switch?

    So again, nVidia is depending on Microsoft to do something for ARM that seems to me a little unlikely. MS saying they’re going to support ARM to enable low-end netbooks/tablets and keep from losing marketshare to Google and Apple is one thing. Expecting this ARM CPU to replace x86 CPU’s, well nah. I doubt it.

  76. I want to know the performance characteristics of both the CPU and GPU.

    Are they trying to get enthusiasts to use it instead of an x86 Windows machine?
    Did they build a better hardware platform for ChromeOS?
    Do they want supercomputers to be built with this instead of x86 CPUs + Tesla?

  77. Could be interesting on the Linux side though. Practically everything in the Linux world works just fine on ARM.

  78. And so now we see the downside to Intel of screwing NV over on chipsets. Depending on how OS and developer tool support plays out, this could be at least as serious a threat to x86 as PowerPC — possibly much more so.

  79. I’m struggling to get out of meh territory with this. If this is to go after the Phenom II and Core i7, maybe, and running Linux mainly it should be just the case of a recompile to go from one to another generally.

    I would have thought this would be for more mobile devices, which aren’t the market I buy from.

    Then again I never take Jen-Hsun Huang seriously.

    Sandy Bridge is what excites me more than anything else going on in the tech industry.

  80. It’s about time Nvidia dug into this section…I can only imagine the cpu’s will be pretty powerful!

  81. In the past Microsoft used an emulator [url=!32<]FX!32[/url<] that allowed x86 Win32 programs to execute on Alpha-based systems running Windows NT

  82. I think this is a great way to go about it, I mean, it [b<]WOULD NOT[/b<] help Nvidia to acquire Via, and Intel and AMD would rather partner up than deal with nVidia again... Didn't Nvidia either acquire or license some of Transmeta's IP, specifically the Code Morphing tech? Could something like 'Project Denver' accelerate APEX or CUDA? That would really point to the "'fully integrated on the same chip' as an Nvidia GPU"

  83. Copyrighted? How? Rosetta was based on licensed code from another company:

    [url<][/url<] [url<][/url<] Maybe you're thinking of the 68k to PPC translator: [url<][/url<] However, that tech went in other virtualization products that didn't get sued out of existence, either, so there's a history of Apple not defending their copyright, if any would apply, that would be taken into consideration by any court of law. That's one of the reasons companies sue so much for copyright--if you don't defend it, courts are likely to decide you don't really care and won't uphold it for you.

  84. I’m reminded of this:

    [url<][/url<] That's from before we knew Nvidia was customising the ARM core. Now that we know they can, it wouldn't be hard to add x86 decode like the article suggests.

  85. Not quite zero apps… they have been pushing .NET for a while, and that should be cross-platform. That doesn’t include many apps I’d care about though (and certainly not many games)…

  86. ^^ soooooooo agree with this. I hate to be a ludite today also but…. the cloud is NOT your friend either, i am willing to bet the big seller will be that these light and spiffy arm-cpu’s will power the next generation of resource less computing….

  87. H AHA HAHHAHAAaaaaa….

    This is awesome, talk about a move no one expected to happen. I’m wondering if Chrome OS will be compatitble? With microsoft demoing Win8 for ARM this week this is a huge announcement. This means that I’ll be able to take my favorite windows apps, IE solid works and auto cad right onto this platform in the years to follow. This might well break intels pc monopoly, KUDOS NVIDIA.

  88. Not much of a battle. Windows x86 has a library of millions of apps and hundreds of thousands of peripheral hardware with working drivers. Windows ARM has exactly zero. Back when MS ruled developers (and knew what developers wanted) they’d have a shot at getting that changed by the time their new OS was out the door. Not so much anymore.

    ARM has a shot at the desktop, but if you were going to bet on an OS it will be running you’d have to pick iOS. With Apple controlling all the apps and delivering them via iTunes, of course. All those people who raged against MS for its supposed iron-fisted control of the Wintel universe are going to discover the meaning of “Be careful what you wish for”

  89. With Windows coming to ARM, this to me finally signals the beginning of CPU-agnosticism. My only question is whether nVidia will develop proprietary binaries to run emulated x86 code a la DEC and its [url=!32<]FX!32[/url<] for the Alpha.

  90. Since apple will have copyrighted-to-death anything close to that or their precious mach core, it will take a pile of coding to fix. HOWEVER! do remember that Microsoft said that windows-8 will be a very “modular” cloud heavy OS… so guess what? they could just run all the heavy hitting on giant server farms and render the result to the device instead. all that would matter at that time would be a frame work for th UI, and transport support to get the data from device to the cloud and back in a efficient manner. that would not be hard at all.

  91. OK if redmond stay out of this I’m fan !

    Could somebody translate fully custom-compatible thing a little bit more ?

    Thanks :]

  92. You do realize it will be at least 2012 for eve a beta yes? it is not so simple to code over supporting binaries that desktop use would need. there are alot of complex development issues to overcome. Given the deep pockets of M$, i do not doubt they will succeed… but don’t hope for a nice shiny win7-ish UI for a while.

  93. Doesn’t say anything about how they’ll run x86 binaries on ARM and vice versa, if that will even be possible.

    They really need something like Rosetta if this is to go anywhere.

  94. I’d be really surprised if one could take a generic Denver system and install iOS on it. Even if Apple’s not got it locked down to require their hardware, it’d take quite a bit of work.

  95. Wow, didn’t see this one coming… Sounds pretty cool though. For the sake of continued competition in the graphics market alone, I hope this does well for them.

  96. I knew it! This is great news!

    Now with the imminent announcement by Microsoft for ARM compatible Windows, the stage is set for Nvidia to battle AMD and Intel.

    Things are looking very exciting for Hardware in 2011 ….drool…

  97. Microsoft’s announcement already came [url<][/url<]

  98. With Android’s momentum, seems to be the right time to this. Let’s see how this will develop.

  99. They really have no choice; with Intel and AMD building graphics into thier CPUS, and ARM adding graphics to their CPUs, Nvidia was probably only a year or two away from being left with a market too small to sustain any version of their current selves. But their core competancies have been graphics and politics, neither of which add much to a server CPU, and I just don’t see a large enough market anytime soon for ARM-based desktop systems – especially ones with a need for very high-end graphics. Wasn’t ‘three years from now’ mentioned as the horizon for delivery of an ARM-based version of Windows?

    It’ll be interesting to see what newness they can bring to the table.