Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
Nvidia working on a processor
Today, Nvidia announced that they're working on an ARM processor. That's right, a CPU, not just a video card. And furthermore, it's going to be for desktops and laptops, not just phones and tablets like Tegra. It's called Project Denver. And it will even be able to run Windows, as apparently Microsoft said that Windows 8 should launch next year and support ARM, in addition to x86.
My guess is that it's meant to be competition for AMD's Bobcat and Via's Nano. ARM is designed for low power consumption, not for high performance. I don't think ARM can be scaled up to offer performance in the same ballpark as Stars or Penryn cores, let alone Sandy Bridge or Bulldozer. Nvidia's graphics are certainly good enough to make for a nice system on a chip for use in cheap laptops.
I'm not sure how wide application compatibility will be. I don't know how plausible it would be to run existing Windows programs on an ARM processor, even if Windows 8 does support ARM. Porting applications should be possible but takes a lot of work from whoever made the program. I'm not sure how hard it would be to recompile code to run on ARM. Emulation would surely be really slow. If it's a mostly disjoint set of applications that run on Windows for ARM versus Windows for x86, that could be a mess, and break the longstanding advantage of Windows that everything would just run. Asus tried an ARM-based Linux "smartbook" and it flopped, badly.
Furthermore, I'm not sure how much of a demand there would be for an Nvidia ARM processor in a desktop or laptop. Any programs that assume good single threaded performance (which means basically everything) would not run well. At best, they'll be competing against AMD's Bobcat and Via's Nano, both of which should be compelling products in the market that Nvidia wants to target--and offer 100% full x86 compatibility, which is something that Nvidia can't promise.
By the time Project Denver launches, it will probably be up against Krishna, not Zacate, and a second generation quad core Bobcat with Caicos integrated graphics on a 28 nm HKMG process will be tough for ARM to catch. Nvidia can probably match the graphics performance if they can get to 28 nm in time, though performance per watt is a different matter entirely unless they're bringing a new GPU architecture to Project Denver. Scaling up ARM cores will be hard, though, and Nvidia has zero experience with CPUs. I don't see Project Denver being a worthwhile product in either desktops or laptops.
Nvidia also wants to target high performance computing (supercomputers) with Project Denver. Intel doesn't see a market there for Atom, nor does AMD see a market there for Bobcat. I don't know what Nvidia sees, other than perhaps most of their traditional markets disappearing.
Comments
since AMD stole much of their GPU business with the 5k series nvidia is trying to find other footholds. however its too little too late. especially the mobile market once the second gen APU hits.
I am rooting for them to have a big breakthrough. I don't particularly like Nvidia, considering their behavior when they almost had the graphics market cornered, but if they disappear, we are going to get gouged by AMD/ATI.
More competition is always a good thing.
East Carolina University, Computer Science BS, 2011
--------------------
Current game: DAOC
Games played and quit: L2, PlanetSide, RF Online, GuildWars, SWG, COH/COV, Vanguard, LOTRO, WoW, WW2 Online, FFXI, Auto-Assault, EVE Online, ShadowBane, RYL, Rappelz, Last Chaos, Myst Online, POTBS, EQ2, Warhammer Online, AoC, Aion, Champions Online, Star Trek Online, Allods, Darkfall.
Waiting on: Earthrise
Names: Citio, Goldie, Sportacus
Intel is doing something similar.
http://kotaku.com/5726127/half+life-devs-say-new-processor-will-bring-console+like-experience-to-pcs
WTF Windows 8? I just got win7 a year ago. Not going to pay more $ for the upgrade.. its cool to see more options on processors.
Edit....... Damn my skimming... lol
Except that Sandy Bridge's integrated GPU is hardly a one-size-fits-all solution to anything. It's a completely underpowered chip for any kind of serious gaming, and buying a PC with one is no different from having any other underpowered GPU.
If someone wants a "console-like experience" on the PC, then get a cheap PC with a low end dedicated graphics chip, bottom out the graphical settings on every game (which you'll have to do on SB's GPU anyways), and there you go. You have console level graphics, GUARANTEED to run at those settings, just like on a console
"since AMD stole much of their GPU business with the 5k series nvidia is trying to find other footholds."
Nvidia has known that much of their market is disappearing for years now. CUDA/GPGPU and Tegra were their initial plans for new markets to keep the company relevant. Project Denver might well be tremendously useful in tablets or portable gaming systems (what do they call Game Boy type devices now, anyway?). But I don't see it having much of a point in desktops.
"Intel is doing something similar."
No, Intel is not doing anything remotely similar. Sandy Bridge is a high performance x86 processor, not a low power ARM processor. x86 is what desktops and laptops have traditionally run on. Indeed, Intel is trying to push x86 even into traditional ARM strongholds such as cell phones.
Intel wants x86 to take over the world, not ARM. If x86 wins, then Intel is one of only three companies that can build the chips--and they get considerable royalties from whatever Via builds. If ARM wins and x86 dies, then ARM Holdings is the big winner, and Intel is just one CPU vendor out of dozens. Intel's only real advantages would be starting with massive amounts of money and being able to get to new process nodes before anybody else. Now, those are certainly substantial advantages, but nothing like the advantages of decades of experience and all but two competitors being legally barred from the market as they have with x86.
It's not at all clear whether ARM can be adapted to the high performance demands of desktops, or whether x86 can be adapted to the ultra low power requirements of cell phones. If they cannot, or perhaps rather, if doing so is so difficult as to not be worthwhile, then ARM and x86 will coexist for the foreseeable future, fighting only at the boundary between low power and high performance needs, which perhaps means tablets and netbooks.
But one thing to remember about Nvidia and ARM is that Nvidia doesn't have any real advantages at all in this market, and it's fiercely competitive. Nvidia doesn't have the same sort of experience with ARM that Qualcomm or TI do. Nvidia does have high performance graphics, but traditional GeForce architectures aren't suitable for the extreme low power situations where ARM makes sense. A low power, low performance CPU paired with a high power, high performance GPU doesn't make a whole lot of sense. The only market that Nvidia dominates that they can use to fund their adventures elsewhere is the professional graphics (Quadro) market, and it's unclear how long their dominance there will last.
"WTF Windows 8? I just got win7 a year ago."
Windows 7 launched in 2009. The claim is that Windows 8 will launch in 2012. Is it really such a shock that Microsoft would release a new OS after three years? The only time I can think of that they went longer than that without a major new release was the XP to Vista transition, and that was because Vista was delayed.
Good, AMD and Intel could use some competition. If it works out Nvidia might actually start making real desktop processors as well.
With just 2 companies making everything stuff is more expensive and I always had the feeling that they sometimes wait to release a new processor because the company already have the best right now and want to sell out the old one first before releasing a new.
the harder the competition the better for the consumers, here we might see some laptops drop in price at least.
I just wish a third company started to make good GFX cards, I miss 3DFX and Matrox (well, Matrox really sucked at the end but they did a few great cards like the Millennium card, still have a old PCI one lying around for when I try to find problems on my buddies computers).
Only reason for nvidia making AMR is because Nvidia doesn’t have X86 license and Intel won’t give them one
AMR has its own values but it’s not really competition to the x86
Totaly wrong man Sandy Bride is Intels Newest x86 CPU with integreated GPU in the same same core
That is old news Quizzical. Nvidia has been doing closed-door development on ARM and other processor projects for years. In fact it tested its first attempt in the server-world in front of many in 2008. However, nothing close to warrant something as mainstream as Intel, who actually has the foundries to really make things work. Just to clue you into the kind of person I am, I've been to many of the live closed-door industry experiments conducted in the United States. Nvidia started in 2003 to give more roles to their own video card processors outside the realm of graphics, due to how development platforms work. ATI started in late 2005 and later being bought by AMD forced them to change their plans, but now that they are owned by a processor company, Nvidia has worked a lot closer for years.
I've done the same thing myself in programming...Where I work on a Prefab Mapping System for two years and then announce to the world a new map pack and developer toolset in development, just to tell the world I am not dead in some communities. Its a very common thing within programming and the tech industry. Shows the world you are still alive and tells the world what you have been doing in silence since us programmers can become very isolated to the world in our work.
I don't have time to post every day in these forums since I actually work within two major industries. I am glad to see however the same old same old going on here. Happy New Year to everyone here. *waves*
note: I love how an announcement occurs, and people take it, post it in every known forumboard...reworded to sound like they are smart and know what they are talking about. Marketting and fans do this all the time. I wish programmers and developers could release articles themselves instead of marketting. You see....when ID and EPIC released their major information, it was through Tim Sweeney (The founder of Epic Games and creator of the Unreal Engine) or Carmack (Founder of ID and creator of the Quake Engine, also authored some graphics optimizations for Nvidia for its doom series of games) who directly wrote PDFs, presented the technologies and documentations so that kids couldn't just reword documents and plaster them across every web page possible.
It's id not ID.
It's even worse than that. AMD stopped being truly competitive in the mid-high (basically CPUs around 150-200+) ever since the Core Duo days. Doesn't help that Sandy Bridge is an absolute monster AND only a midstream product. I love my Intel CPUs, but someone really needs to challenge them so I stop getting bent over everytime I upgrade.
Alltern8 Blog | Star Wars Space Combat and The Old Republic | Cryptic Studios - A Pre Post-Mortem | Klingon Preview, STO's Monster Play
That Nvidia is working on an ARM processor isn't news in itself. They've been shopping Tegra for smartphones and perhaps tablets for quite some time, without much success. That they've got ideas about pushing ARM into desktops and HPC is news, however--especially when combined with Microsoft's essentially simultaneous announcement that Windows 8 will support ARM.
The desktop side of this makes me wonder what Nvidia has in mind. Are they going to push for a replacement for the now all-but-dead Atom+Ion combination? Or are they going to try to make ARM compete with x86 in performance?
-----
Intel dominates the high end right now, but with the arrival of Bulldozer in the spring, that could change. Sandy Bridge beats Phenom II in single threaded performance by about 30% or so, and it's really a question of how much of that gap Bulldozer can make up.
If Bulldozer manages to beat Sandy Bridge in single-threaded performance, it could easily turn into AMD dominating the high end. Even if Bulldozer narrowly loses in single-threaded performance, having more cores and superior chipsets will still make it a compelling platform. If Bulldozer is barely able to narrow the gap at all, then AMD would be stuck in their current position of trying to offer more cores for cheaper, and Intel will dominate the high end of the market for years to come.