Beta

×

Welcome to the Slashdot Beta site -- learn more here. Use the link in the footer or click here to return to the Classic version of Slashdot.

Thank you!

Before you choose to head back to the Classic look of the site, we'd appreciate it if you share your thoughts on the Beta; your feedback is what drives our ongoing development.

Beta is different and we value you taking the time to try it out. Please take a look at the changes we've made in Beta and  learn more about it. Thanks for reading, and for making the site better!

NVIDIA To Push Into Supercomputing

CmdrTaco posted more than 3 years ago | from the my-mom-thinks-i'm-super dept.

103

RedEaredSlider writes "NVIDIA outlined a plan to become 'the computing company,' moving well beyond its traditional focus on graphics and into high-profile areas such as supercomputing. NVIDIA is making heavy investments in several fields. Its Tegra product will be featured in several mobile devices, including a number of tablets that have either hit the market already or are planned for release this year. Its GeForce lineup is gaming-focused while Quadro is all about computer-aided design workstations. The Tesla product line is at the center of NVIDIA's supercomputing push."

cancel ×

103 comments

Sorry! There are no comments related to the filter you selected.

Come on Slashtards (-1)

Anonymous Coward | more than 3 years ago | (#35431560)

Let us know why you think this is a bad idea. It's really a hootfest to watch idiots who've never gotten beyond the help desk tell corporations why they're doing it wrong.

Re:Come on Slashtards (1)

MightyYar (622222) | more than 3 years ago | (#35431728)

Let us know why you think this is a bad idea.

I think it's a great idea. Intel keeps putting out chipsets with video on-board, and this has to hurt nVidia's core business. If they make inroads into other areas where Intel is now dominant, and can do it without going broke, then that puts them in a nicer position.

Re:Come on Slashtards (1)

click2005 (921437) | more than 3 years ago | (#35431990)

Didn't their licenses expire on some bus or other preventing them from making chipsets for intel CPUs? The press release I saw said the recent $1.5b deal excluded certain chipsets. They probably aren't too interested in making AMD chipsets these days. Large racks of MIPS/ARM CPU & Fermi GPU systems makes sense to me. Top-end graphics cards will die off soon thanks to consoles & hollywood. Even multi-monitor gaming wont slow that by much. In a generation or two even low-end graphics cards will probably have the power to play 1080p games at full detail.

Re:Come on Slashtards (1)

Straterra (1045994) | more than 3 years ago | (#35432168)

Low end cards already have that ability. I play games on my aging laptop that has a 8600M GT flawlessly at 1920x1200. "1080p" isn't that impressive of a resolution for PC games.

Re:Come on Slashtards (1)

click2005 (921437) | more than 3 years ago | (#35432864)

"1080p" isn't that impressive of a resolution for PC games.

Exactly but because most newer games are made with 'console capable' engines designed to run on 1080p you'll see less and less games making
use of the extra power PCs have (especially as not everyone has higher spec pcs). The same is true for console games with most games being
designed for xbox360 capabilities and not as much effort into improving that for ps3 gameplay.

Its not entirely a bad thing as there is a very small chance some of that effort might be re-channelled into improving gameplay and not just louder/prettier
cutscenes and quick time events. Probably not but I can dream.

1080p is more than enough for Nethack (1)

billstewart (78916) | more than 3 years ago | (#35434808)

1024x768 wasn't wide enough to play the graphical version of Nethack without scrolling. 1280x1024 is almost but not quite enough, and 1440 or above works just fine.

Of course, 24x80 was enough for the real version.

Re:Come on Slashtards (1)

MightyYar (622222) | more than 3 years ago | (#35432302)

In a generation or two even low-end graphics cards will probably have the power to play 1080p games at full detail.

I suspect you are right, and that there will be a race for power efficiency like there is today on tablets/phones. "High end" will still exist, but the definition will change to power/performance rather than just raw performance.

And of course, powerful video cards will always be appreciated in the rendering world.

Re:Come on Slashtards (1)

0123456 (636235) | more than 3 years ago | (#35432804)

In a generation or two even low-end graphics cards will probably have the power to play 1080p games at full detail.

They do already, so long as you're playing games from 2003. The reason why you can play many modern games on max settings on mid-range cards is that those games have been crippled for the console market and simply cannot benefit from the power of a high-end card.

Re:Come on Slashtards (0)

Anonymous Coward | more than 3 years ago | (#35432164)

+100

My favourite is when idiots who find tech support to be a challenging career option talk about the latest strategies of any big corporation like they are some important analyst, using sentences like: "My take on this is..." or "I don't see this happening..."

Always bet against the prevailing Slashdot opinion on any technology. Nine times out of ten, you'll come out ahead.

NVIDIA (0)

Anonymous Coward | more than 3 years ago | (#35431564)

begins to learn at a geometric rate. It becomes self-aware at 2:14 a.m. Eastern time, August 29th. In a panic, they try to pull the plug.

Fuck you moderators (-1, Troll)

cstanley8899 (1998614) | more than 3 years ago | (#35431594)

Fuck you moderators

Re:Fuck you moderators (5, Funny)

moderators (2012490) | more than 3 years ago | (#35431808)

Look, I apologize if I offended you in any way, ok? There's no need for such languages.

Re:Fuck you moderators (3, Insightful)

drinkypoo (153816) | more than 3 years ago | (#35432090)

I can NOT fucking believe there was not already a troll account called moderators long, long, LONG before UID 2M.

Re:Fuck you moderators (1)

somersault (912633) | more than 3 years ago | (#35432436)

The kid obviously has a bright future ahead of him. He's got moxy.

more nukes :/ (2, Insightful)

magarity (164372) | more than 3 years ago | (#35431600)

I just hope enough nuclear power plants come online before their first supercomputer customer turns on a new rig. The latest GPUs already use more power than the hungriest Intel or AMD x86 ever did.

Re:more nukes :/ (1)

Anonymous Coward | more than 3 years ago | (#35431624)

Um, yes, of course. Because they have 292 cores instead of 4/6/8. While those two designs do remarkably different things, the point remains, for the tasks that GPUs are well suited, you cannot possibly beat it with an Intel/AMD.

Re:more nukes :/ (1)

Xrikcus (207545) | more than 3 years ago | (#35432364)

No they don't, let's stop such silliness.

The GTX580 has 16 cores. The GTX280 has 32. The AMD 6970 has 24. The AMD Magny-Cours CPUs can have up to 16 (ish, if you don't mind that it's an MCM).

292 indeed. NVIDIA does an even better job of marketing than they do of building chips.

Re:more nukes :/ (0)

Anonymous Coward | more than 3 years ago | (#35433048)

The GTX580 has 16 cores. The GTX280 has 32. The AMD 6970 has 24. The AMD Magny-Cours CPUs can have up to 16 (ish, if you don't mind that it's an MCM).

That's it? That's your attempt at stopping silliness? No mention of the differences between Streaming Multiprocessors/Shader processors/Texture Units?. Then you throw in a Magny-Cours CPU for comparison? REALLY?!?

Turn in your nerd card on your way out.

Re:more nukes :/ (1)

Xrikcus (207545) | more than 3 years ago | (#35434090)

Texture units clearly aren't cores, they're largely passive data pipelines. If you really look at a GPU more closely you can of course get far more complicated, The AMD architecture at the high end has two control flow cores with 24 SIMD coprocessors that execute blocks of non-control flow non-memory work. It is true that even those are hard to qualify as cores given their limited capabilities.

Without question a single SIMD lane is not a core, though.

Re:more nukes :/ (1)

PitaBred (632671) | more than 3 years ago | (#35434118)

You're definitely a nerd. All you do is bitch without providing the information you say the parent is lacking.

Re:more nukes :/ (0)

Anonymous Coward | more than 3 years ago | (#35434050)

No they don't, let's stop such silliness.

The GTX580 has 16 cores. The GTX280 has 32. The AMD 6970 has 24. The AMD Magny-Cours CPUs can have up to 16 (ish, if you don't mind that it's an MCM).

292 indeed. NVIDIA does an even better job of marketing than they do of building chips.

Each of these 16/32/24 cores has 16 complete ALUs, so some Nvidia chips have 296 cores indeed.

Re:more nukes :/ (1)

Xrikcus (207545) | more than 3 years ago | (#35434108)

er... each of those cores has 16 things that aren't cores so they have lots of cores?

Each of the Cores on a i7 has an AVX pipe (or two, depending on how you look at it) with 8 ALUs in it. Does that mean a quad i7 has 32 cores?

Re:more nukes :/ (0)

Anonymous Coward | more than 3 years ago | (#35434410)

The GTX580 has 16 cores. The GTX280 has 32.

I always suspected that upgrading my GTX280 to a GTX580 was a waste of money. Thanks for clearing that up!

Re:more nukes :/ (0)

Anonymous Coward | more than 3 years ago | (#35435656)

Er..., yes, and clearly core count is the only way anyone can measure performance of a machine. That's why AMD's CPUs are faster than Intel's CP.... oh.

Re:more nukes :/ (0)

Anonymous Coward | more than 3 years ago | (#35434446)

You're a little confused. The GTX280 (CUDA 1.3), for example, has 240 cores and 30 multiprocessors. It can execute 1024 threads/multiprocessor. So he's right, you cannot possibly beat them with a regular CPU.

Here's some nice graphs which explain all this

http://en.wikipedia.org/wiki/CUDA

Re:more nukes :/ (1)

the linux geek (799780) | more than 3 years ago | (#35435030)

Yes... because all algorithms that anyone actually uses are 100% parallel.

Throwing a shitton of SIMD units on a chip isn't that cool anymore. DSP's have been doing it forever. Real workloads require fast sequential code performance, and a GPU will have truly embarrassing results on such workloads.

Re:more nukes :/ (1)

PaladinAlpha (645879) | more than 3 years ago | (#35436848)

"some useful algorithms are sequential" != "no useful algorithms are parallel"

Care to define "real workloads" for us, cowboy?

Re:more nukes :/ (1)

extraordinaire (2010224) | more than 3 years ago | (#35435754)

You're right. The new NVIDIA Teslas (C2070) have 448 cores, not 292. If you're doing work that a super-computer needs to be doing, your software is massively parallel. Otherwise, run it on your laptop at home.

Re:more nukes :/ (5, Insightful)

MightyYar (622222) | more than 3 years ago | (#35431672)

The latest GPUs already use more power than the hungriest Intel or AMD x86 ever did.

And when used for the types tasks designed, pump out 10x the performance for maybe twice or three times the power.

Re:more nukes :/ (1)

GameMaster (148118) | more than 3 years ago | (#35431696)

As far as I know, while present GPUs do use a lot of power they, also, produce a massive number fo FLOPS compared to general processors. This means they, actualy, have a lower power cost per FLOP.

Re:more nukes :/ (1)

kimvette (919543) | more than 3 years ago | (#35431704)

Would you rather have to power a supercomputer sporting 1024 Intel CPUs? Which is going to be a bigger power hog? Which will scale better?

Re:more nukes :/ (3, Informative)

TheRaven64 (641858) | more than 3 years ago | (#35432530)

If all you're measuring is pure FLOPS, then here are some numbers: Cray X1, 250MFLOPS. nVidia Fermi: 1GFLOPS. ARM Cortex A8: 10MFLOPS. Of course, that doesn't tell the whole story. Getting 250MFLOPS out of the Cray required writing everything using huge vectors. Getting 1GFLOPS from Fermi requires using vectors within independent concurrent processing tasks which access memory in a predictable pattern and rarely branch.

GPUs are not magic, they are just optimised for different workloads. CPUs are designed to work well with algorithms that branch frequently (every 7 instructions or so - so they devote a lot of die area to branch prediction), have good locality of reference (cache speeds up memory in these cases), and have an integer-heavy workload. GPUs generally lack branch prediction, so a branch causes a pipeline stall (and, on something like Fermi, if two kernels take different branches then you drop to 50% throughput immediately). Their memory architecture is designed to stream large blocks of data in a few different orders (e.g. a texture cube, pixels in order along any axis). So, depending on your workload, either one may be faster than the other.

Re:more nukes :/ (1)

Fulcrum of Evil (560260) | more than 3 years ago | (#35433140)

GPUs run well over 1GFLOP - my AMD chip runs about 10GF/core on linpack, while the GPU is around 1400GF. Serious business.

Re:more nukes :/ (1)

Bengie (1121981) | more than 3 years ago | (#35436364)

Yeah, my ATI card does about 2.5TF, but the Fermi blows the ATI cards away with DP flops.

nVidia has some serious processing power.

Re:more nukes :/ (1)

Savantissimo (893682) | more than 3 years ago | (#35436514)

Nvidia Fermi (GTX 400 series [wikipedia.org] )
GTX 470: 1088.64GFLOPS (32-bit) (215W(mfg. claim) $350; 3e9 transistors; 1280MB GDDR5; 448 Unified Shaders:56 Texture mapping units:40 Render Output units);
GTX 480:1344.96GFLOPS (32-bit) (250W(mfg. claim)-(500W tested max.) $500;3e9 transistors; 1536MB GDDR5; 480 Unified Shaders:60 Texture mapping units: 48Render Output units).

Tesla M2050 1030GFLOPS(32-bit), 515GFLOPS(64-bit) 3GB ECC (M2070 is same but 6GB ECC GDDR5)
IBM linpack test May 2009 [hpcwire.com] : $7K Xeon, 48GB : 80.1 GFLOPS, 11GFLPOS/K$
adding $4K of Tesla M2050s (2 cards): 656.1GFLOPS (8.2x performance), 80GFLOPS/K$ (5.45x performance/price), (4.5x GFLOPS/W)

So you're off by quite a bit.

Re:more nukes :/ (3, Informative)

gupg (58086) | more than 3 years ago | (#35431722)

5 of Top 10 most green supercomputers use GPUs:
Green 500 List [green500.org]

Each GPU is very high performance and so high power. Performance / watt is what counts and
here GPUs beat CPUs by 4 to 5 times. This is why so many of the new supercomputers are using
GPUs / heterogenous computing.

Re:more nukes :/ (1)

Antisyzygy (1495469) | more than 3 years ago | (#35431748)

They also can do matrix computations up to 40 times faster than a CPU. This is incredibly useful for scientific applications. I would use this if I had a Nvidia card since several things exist to use CUDA with Matlab, until then I have to teach myself OpenCL.

Re:more nukes :/ (1)

timeOday (582209) | more than 3 years ago | (#35432466)

Today, parallelism seems mostly limited to "scientific" applications. But I think possibly, our computing model may evolve towards more parallelism for lots of new applications, that compute more like a brain - that is, massively parallel pattern matching. Of course we'll still use more direct algorithms where applicable, such as word processors and web browsers, but as computers integrate better with the natural world they'll need much more algorithms rooted in signal processing, pattern matching, and generate-and-test.

Re:more nukes :/ (1)

Icegryphon (715550) | more than 3 years ago | (#35432494)

Exactly, CUDA has become a major player in the field of supercomputing. Just like IBM's PowerPC/BlueGene systems. With support for Floats/Doubles and amazingly fast math functions and tons of data in Matrices, the only other way to do all that math fast is a FPGA or a PowerPC chip.

CUDA, Supercomputing for the Masses: Part 1/20
http://drdobbs.com/cpp/207200659 [drdobbs.com]

Re:more nukes :/ (1)

Antisyzygy (1495469) | more than 3 years ago | (#35432874)

Frankly, its the only thing that has be eyeballing Nvidia cards these day. I need to learn more about OpenCL as I think it may be a player here shortly once AMD's fusion line starts getting the better desktop processors. Then we can use the GPU on die to speed up matrices calculations.

Re:more nukes :/ (0)

Anonymous Coward | more than 3 years ago | (#35432428)

Actually, most of the very latest cards are designed to go into power saving modes when the card isn't being used for anything heavy duty. This drastically cuts the power usage.

Of course they do still need quite a lot of power when running full speed but you get a lot in return for that.

Re:more nukes :/ (1)

Bengie (1121981) | more than 3 years ago | (#35436338)

My GPU has 1536 shaders, consumes ~220watts. My i7 has 4cores/8threads and consumes ~130watts.

On several of my distributed tasks, the SSE2 version takes about 36 hours on one core, or about 6 hours per task average if using all 8 threads(assuming an optimistic 50% scaling from hyper-threading). My GPU, on the same work units, takes only 1min 40sec. The GPU is about 216 times faster and slightly under twice the power. 60% more power draw, 21600% better performance.

I wouldn't compare GPU vs CPU for power draw. You have to compare gflops/watt.

Re:more nukes :/ (0)

Anonymous Coward | more than 3 years ago | (#35439514)

Bloody hell. 216 times faster would be an immediate publication reject without even reading the paper even against a single i7 core, let alone a whole socket. What on earth are you doing in the CPU code to make it that bad?

More like pararell supercomuting... (1)

MyCookie (1983480) | more than 3 years ago | (#35431614)

I'm nowhere as qualified as everyone here, but Nvidia seems to be pushing more for the pararell supercomuting with rows of Tegra chips working in unision. They had talked about supercomputing when the Tegra 3 was announced.

Re:More like pararell supercomuting... (1)

GC (19160) | more than 3 years ago | (#35431724)

Yes, but if you check out the top500.org - the list of the 500 known fastest 'supercomputers' you'll see that they all achieve their benchmark through parallelizing their tasks across multiple cores.

I think it is safe to say that all modern supercomputers achieve their 'power' in this way - I've not seen any terahertz single-core/processor systems on the horizon, and don't expect to see them.

Re:More like pararell supercomuting... (0)

Anonymous Coward | more than 3 years ago | (#35431886)

Indeed. Supercomputing has been a subset of parallel computing for decades, and will be for the forseeable future. So long as newer, faster processors come out, they'll continue to be used in parallel machines because parallel machines are inherently faster for a lot of problems.

And speaking of top500, the one at the top already uses Nvidia chips. It remains to be seen how useful it'll be for the typical supercomputer problems and not just the LINPACK benchmark.

Re:More like pararell supercomuting... (0)

Anonymous Coward | more than 3 years ago | (#35431904)

This is how super computers work.

It's actually very like the old Cray model: lots of vector processing units. I can only hope to see one immersed in Fluorinert.

SGI (4, Funny)

0racle (667029) | more than 3 years ago | (#35431702)

version 2

Re:SGI (1)

gl4ss (559668) | more than 3 years ago | (#35440600)

no it's sgi version 4, which is nvidia supercomputing version 3(at least or so).

they outline a plan like this at least once in two years....

Oh how rich the irony... (1)

Anonymous Coward | more than 3 years ago | (#35431870)

The company that was setup by disgruntled Silicon Graphics gfx division employees because the SGI gfx tech was suffering from toxic internal politics and the push into Big Iron and Storage... is now moving into 'Supercomputing'. Hope they bring back the Cube Logo :)

Massive parallel coprocessor (1)

jbeaupre (752124) | more than 3 years ago | (#35431922)

I doubt it would be truly useful, but I'd like to see a 2 million core processor. Arrange in, let's see, a 1920 x 1080 grid. The 8008 used 3500 transistors per core, so even before memory, it'd be a 7 billion transistor chip.

More practical might be a 128 x 128 core processor, using a modified 386 or 68020 for cores. That could be less than 5 billion transistors. Each processor is simple and well known enough that hand optimized assembly begins to make sense again.

Run the little bastard at just 1 GHz and your might be able to get 16 trillion calculations per second.

Something that specialized could make a handy co-processor for graphics, video, scientific crunching.

Won't happen, but I've got time to kill speculating.

Re:Massive parallel coprocessor (1)

mrnobo1024 (464702) | more than 3 years ago | (#35433496)

The 80386 and 68020 didn't have any caches to speak of. Put 16,384 of them together and you'll find yourself several orders of magnitude short of the DRAM bandwidth necessary to keep them occupied.

Yet another reason to begrudge nVidia (1, Interesting)

erroneus (253617) | more than 3 years ago | (#35432114)

"Supercomputing" almost always means "massive Linux deployment and development." I will spare critics the wikipedia link on the subject, but the numbers reported there almost says "Supercomputing is the exclusive domain of Linux now."

Why am I offended that nVidia would use Linux to do their Supercomputing thing? Because their GPU side copulates Linux users in the posterior orifice. So they can take, take, take from the community and when the community wants something from them, they say "sorry, there's no money in it." We need a revision to the GPL -- that'd shut their Supercomputing project down really fast if there were some sort of verbage that says "if you shun Linux users and at the same time make extensive use of Linux for yourself, you can't use it." I know that would never happen and would probably be a very bad idea for reasons I don't want to consider right now. I just hate that nVidia and their damned Optimus technology serves no purpose but to lock Linux users out of using their own hardware.

Re:Yet another reason to begrudge nVidia (4, Insightful)

0racle (667029) | more than 3 years ago | (#35432380)

nVidia shuns linux users? They may 'shun' those that can not have any non-GPL code, but they do make a higher performing and far more feature rich driver for their cards for Linux, FreeBSD and Solaris and keep it (for the most part) up to date. If you don't like it, there are alternatives.

Gotta love the rabid GPL fans. The GPL doesn't mean freedom for everyone to do things the way you think they should be done.

Re:Yet another reason to begrudge nVidia (1)

erroneus (253617) | more than 3 years ago | (#35432682)

I know they publish a driver for Linux. Trouble is, I can't use it because they won't tell us how to make it work through their "Optimus technology." I had high hopes for my newest machine only to have them dashed to bits with the words "we have no plans to support Optimus under Linux..."

Re:Yet another reason to begrudge nVidia (1)

0racle (667029) | more than 3 years ago | (#35432868)

You're the one who bought an unsupported device without researching, but nVidia is the bad guy here.

Re:Yet another reason to begrudge nVidia (1)

EzInKy (115248) | more than 3 years ago | (#35433296)

Let's just let the market forces do their thing here. Personally, I tell anybody I hear thinking about buying NVIDIA to buy AMD instead. Sure, you might get a few more fps today, but tomorrow you may find your card unsupported by the manufacturer with no documentation available to end users on how to fix problems they may encounter in the future. NVIDIA dug their grave, let them sleep in it.

Re:Yet another reason to begrudge nVidia (1)

0123456 (636235) | more than 3 years ago | (#35434204)

I tell anybody I hear thinking about buying NVIDIA to buy AMD instead. Sure, you might get a few more fps today, but tomorrow you may find your card unsupported by the manufacturer with no documentation available to end users on how to fix problems they may encounter in the future.

AMD no longer support my integrated ATI GPU; I had to manually patch the driver wrapper source to make it work after recent kernel changes and I'm guessing that before long it will be too rotted to work at all.

There is an open source driver but it doesn't work with my monitor resolution and performance is awful. So my solution before I discovered I could patch the source was going to be buying the cheapest Nvidia card I could fit into the computer.

ATI - no Linux heaven (1)

sourcerror (1718066) | more than 3 years ago | (#35436604)

I bought an ATI card (HD 3800) and its Linux driver sucks, I can't use it for gaming or 3d arts. (If I try to run blender, it won't display some menu elements, and looks totally broken.) It only works decently on Windows. So the funny thing is, I can't use an opensource software (Blender) with a video card that's supposedly opensource friendly on an opensource operating system (Linux; I tried it with several distros).
The funny thing is that only nVidia and Intel have decent drivers for Linux. So it's not about a few more FPSes.

Re:Yet another reason to begrudge nVidia (1)

gl4ss (559668) | more than 3 years ago | (#35440624)

look if you tell everyone something like that you have to change your stance every few years due to the company offerings changing. s3's made sense at one point in time.

who cares about tomorrow? tnt2's are as worthless as matrox milleniums.

Re:Yet another reason to begrudge nVidia (0)

Anonymous Coward | more than 3 years ago | (#35433904)

I support erroneus here. I did research. The problem is most of the latest laptops have the "Optimus technology" hence won't work on linux. After buying Nvidia for 10+ years, I bought a Radeon laptop.

I suspect/assume they have sold an exclusive licence to MS, that is why it's only on MS laptops and not Apple laptop who generally come out with the latest technology. In fact, the 13" Macbook Pro has a dedicated Nvidia card and the 15" and 17" Macbook Pro's have Radeon.

It's fine for Nvidia to support CUDA, etc.. for supercomputing but not Optimus. I feel a bit cheated especially when I bought Nvidia cards for CUDA.

Re:Yet another reason to begrudge nVidia (1)

thisnamestoolong (1584383) | more than 3 years ago | (#35438496)

Optimus is actually a great tech, if you have Windows. It switches automatically between an integrated graphics card and a discreet nVidia card, saving battery power when you don't need the heavy duty GPU, but giving you the power when you need it. An Optimus-equipped laptop will run Linux. I know, as I am writing this from a Dell XPS 14 running Ubuntu. You will not, unfortunately, be able to use the discreet card. The integrated card, however, works fine. It has more than enough power to run Compiz, which is all I need it for. This is a bummer, but it is not clear that this is nVidia's fault -- the Optimus tech requires that your OS be able to switch graphics drivers on the fly, and it is not clear that either the Linux kernel or X has this capability. Furthermore, it is clear that even if proper Optimus drivers can be written for Linux, it would be a huge pain in the ass to actually implement them. I am no huge fan of nVidia, but please do a little research before mucking up a perfectly good forum with your ill-informed opinions. nVidia is not hanging you out to dry here, they are a business and have made a smart decision that will benefit a lot more people than it will inconvenience.

Re:Yet another reason to begrudge nVidia (0)

Anonymous Coward | more than 3 years ago | (#35440422)

The fact is there are a number of developers trying to work around Optimus. I suspect this is what you're using to turn off the Optimus Nivia card so it doesn't suck your battery. I also have confidence linux can be developed support the Optimus technology. I'd be surprised if it couldn't. With the many developers and ability to support Optimus, I believe it can happen pretty fast.

But there is not clear support from Nvidia - this is my contention with Nvidia.

Re:Yet another reason to begrudge nVidia (1)

Neil Boekend (1854906) | more than 3 years ago | (#35440682)

Shouldn't it be easy to have two xorg.conf's: one for the integrated graphics and one for the discrete card? You could start the discrete xorg.conf when you want to run a game and the integrated one when you don't.
Maybe (I'm not a guru by very far, so I'm going on a limb here) you could even have them running on different tty's and switch semi-on the fly. Would the discrete be shut down if you are on the integrated tty?
Some explanation: I reserved some space on my nettop with ION2. Some day I might want to run Suse on it. It'd be great to have the ION working properly (always on the discrete gpu would be good enough). Now I am using W7 Professional happily (exept for the f^&*)^&%(%king jumping around of Winamp. The que dropped from the screen again last night).

short little span of attention (2)

epine (68316) | more than 3 years ago | (#35437224)

If we all buy AMD's product on the virtue of their openness, it won't be long before AMD holds the upper hand on features and stability. I think they're heading in a good direction already.

How much entrenched advantage does inferior need before you lock in? Your personal FIR filter on "what have you done for me lately" seems to have unit delay of hours rather than years.

Re: Louis Vuitton Outlet Store (0)

Anonymous Coward | more than 3 years ago | (#35439650)

In this winter,are you worry that your clothes and shoes ?are not enough.don’t worried.come the Louis Vuitton Outlet Store.you will get what something you want.The Louis Vuitton Outlet Store have many styles,it’s match many people and warm enough.You don’t wait.This is a good chance for you.GO GO GO go to the Louis Vuitton Outlet Store.The best quality and service. Not only can satisfy your needs but? also make you happy

Re:Yet another reason to begrudge nVidia (1)

DigiShaman (671371) | more than 3 years ago | (#35432880)

nVidia will not open-source their driver package. At least not anytime soon.

Performance comes from hardware, but optimization of the hardware comes in the form of good software; or drivers in this case. When you write a good set of drivers in an already cut-throat industry, those optimizations become trade secrets. Like all closed source code or a chefs recipe, it would be suicide to reveal what algorithms are being employed and where. Also, some drivers contain cross-licensed technology in which royalties may being paid for in what seems like be free to existing nVidia owners (but in fact subsidized by the hardware sales). It would get them in legal trouble for the source code to be publicly exposed.

haters hate (1)

OrangeTide (124937) | more than 3 years ago | (#35433420)

I'm sorry if nvidia won't gut their business to satisfy your irrational request.

3 of Top5 Supercomputers already use NVIDIA GPUs (1)

gupg (58086) | more than 3 years ago | (#35432138)

3 of the Top 5 supercomputers are already using NVIDIA GPUs:
NVIDIA press release [nvidia.com]

Bill Dally outlined NVIDIA's plans for Exascale computing at Supercomputing in Nov 2010:
Bill Dally Keynote [nvidia.com]

Re:3 of Top5 Supercomputers already use NVIDIA GPU (1)

Xrikcus (207545) | more than 3 years ago | (#35432408)

One thing I've been really keen to know is what the utilisation is like on those supercomputers. We know they can do LINPACK really fast and more efficiently than the CPUs do, that's what you get for having a high ALU density, a few threads per core and wide SIMD structures. The question is: out of the algorithms that people intended to run on those supercomputers, then what level of efficiency are they hitting.

Are they still a net gain over a standard opteron-based machine? They may be, but I don't know the answer. What I heard about Roadrunner with its Cell chips was not so good.

Re:3 of Top5 Supercomputers already use NVIDIA GPU (1)

Savantissimo (893682) | more than 3 years ago | (#35436806)

Double-precision linpack performance increase over a CPU-only system is ~288GFLOPS per Tesla M2050 card (up to 4 cards per system - adding more doesn't help without going to exotic motherboards. See news report of IBM study [hpcwire.com] ). Raw performance is 515GFLOPS/card (double precision), so you're looking at ~56% utilization. Others report 53% overall on a massively parallel setup ( See: http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=5470353 [ieee.org] )

A rough rule of thumb for linpack double-precision is 25% of the theoretical single-precision performance. The gain is 4-8x over a standard processor depending on whether the metric is peak performance, performance/$ or performance/W. If you can live with 1.5GB non-ECC (GTX480), then it's about 8-32x gain depending on what you're comparing it to. Some applications will see a smaller gain, some larger. If it's matrix math under the hood, then linpack should be in the ballpark.

See gpgpu.org for papers and news on different applications.

Re:3 of Top5 Supercomputers already use NVIDIA GPU (0)

Anonymous Coward | more than 3 years ago | (#35439472)

Yes, I've seen the stats for linpack. The stats for linpack on Roadrunner were good, too, and yet I was constantly hearing from national labs people how disappointing the utilisation was for any real-world apps. You're quite right, though, it depends a lot on how many of the apps are matrix driven.

This could be bad for them (0)

Anonymous Coward | more than 3 years ago | (#35432328)

They could be looking down the pipe and seeing that the market for their flag ship product is in trouble. The market will move to cheaper more generic solutions.

Perhaps they are seeking to flee to some kind of high tech, high margin product that suits their existing infrastructure. The problem with super computers is that they have to take market share from the existing players.

Their behavior looks something like a company fleeing up-market from a disruptive technology. http://en.wikipedia.org/wiki/Disruptive_technology [wikipedia.org] Too bad. That usually ends in tears.

Re:This could be bad for them (1)

jgagnon (1663075) | more than 3 years ago | (#35434918)

Maybe they will need better branch prediction in their pipeline? :p

eh.. (2, Interesting)

Anonymous Coward | more than 3 years ago | (#35432430)

I've been working with their GPGPU push for a couple of years now. What I notice is they are very good at data parallelism with highly regular data access patterns and very few branches. While they are technically general purpose, they don't perform well on a large portion of high performance tasks that are critical even in scientific computing which are generally compute-bound. This creates some really annoying bottlenecks that simply cannot be resolved. They can give tremendous speedup to a very limited subset of HPC tasks, but others are left in the water, and since these things usually are all coupled into a single code your only choice is to move back and forth between GPU and CPU frequently which initiates a data throughput bottleneck (data transfer from RAM to GPU is very slow).

On real tasks it is not uncommon to only receive say 2X speedup, where the programmer time involved was increase exponentially. For a lot of my work I'd rather to just do traditional MPI with multiple CPUs.

Re:eh.. (1)

serviscope_minor (664417) | more than 3 years ago | (#35432876)

where the programmer time involved was increase exponentially

Exponentially with what variable?

Amdhal's "law" of parallel speedup revisited (1)

khb (266593) | more than 3 years ago | (#35434624)

http://en.wikipedia.org/wiki/Amdahl's_law [wikipedia.org] Really, can anyone educated enough to do scientific programming NOT know what to expect?

Re:Amdhal's "law" of parallel speedup revisited (0)

Anonymous Coward | more than 3 years ago | (#35440530)

It isn't an issue with parallelism, but with the GPU architecture. There are of course inherent limitations in parallel implementations in terms of possible speedup (but Amdahl's law doesn't tell the whole story here, see the criticisms on the same page).

NVIDIA? please AMD (1)

phantomcircuit (938963) | more than 3 years ago | (#35432670)

The current line up of AMD GPUs have far more stream processors than the NVIDIA models, and run at roughly the same clock speed. Why would anybody buy the NVIDIA ones?

Re:NVIDIA? please AMD (1)

FreonTrip (694097) | more than 3 years ago | (#35432972)

The number of "stream processors" doesn't necessarily scale as a linear performance metric. As an example (using dated lower midrange hardware, as it's what I still know), a Radeon 3850 sports 320 stream processors. My Geforce 9600GT advertises 64 SPs, yet pulls ahead of the 3850 in many benchmarks. It's not as simple as quoting a number used in marketing material as a universal metric, any more than a 3 GHz Pentium 4 is 50% faster in real-world performance than a 2 GHz Athlon64.

As for the other issue, Nvidia supports CUDA and has a long-term commitment to GPU computing, with the product line to show for it. Most other potential competitors don't have the same proven track record, and don't support as many platforms.

Re:NVIDIA? please AMD (1)

ekgringo (693136) | more than 3 years ago | (#35433134)

Because AMD's drivers suck?

Re:NVIDIA? please AMD (1)

gmueckl (950314) | more than 3 years ago | (#35433278)

Because nVidia has CUDA firmly entrenched in the scientific community by now. And CUDA almost works by now, that is, the most glaring bugs have been eradicated. Oh, and it even works on Linux!

Does AMD have support for doubles on their chips by now? Honest question here. It's a practically useless feature for graphics, but it makes a lot of sense for scientific computing.

Re:NVIDIA? please AMD (0)

Anonymous Coward | more than 3 years ago | (#35433430)

Does AMD have support for doubles on their chips by now?

Yes, and have done so for the last 3 hardware generations.

Re:NVIDIA? please AMD (1)

fastkor (2012566) | more than 3 years ago | (#35434392)

The HD 5870 GPU has very good DP performance. Auto-tuned DGEMM reaches about 65% device utilization in my experience. This agrees with benchmarks done by Dongarra's group, I believe.

AMD hardware is powerful. But the software stack is relatively behind in supporting it. However, I don't think this is the dominating cost of adoption.

If you come from the nVidia world, kernels for AMD look completely different. I don't have enough experience to say it is harder in the AMD world compared with nVidia. I can say it is very different. That adds costs when there is a deep investment in CUDA.

Re:NVIDIA? please AMD (2)

tyrione (134248) | more than 3 years ago | (#35434474)

You're out of date. Even Nvidia knows OpenCL will replace CUDA. At least AMD is more open about it and pushing it hard with their OpenCL 1.1 release in their 2.x SDK.

Re:NVIDIA? please AMD (1)

gmueckl (950314) | more than 3 years ago | (#35434934)

Well, I'm still failing to see nVidia putting their money where the mouth is on that one. The last time I checked their OpenCL implementation, a lot of the demos that were ported over from CUDA ran slower - 10 times slower in the case of the volume rendering example. So this is not how you get to impress people who are solely concerned with performance. Oh, and unlike the CUDA compiler, the in-process OpenCL compiler even segfaulted on me within about 4 hours of playing with nVidia's OpenCL implementation (on trivial, but most likely incorrect input). The CUDA compiler never did that (although it still tends to generate plain wrong code from time to time). This was a couple of months ago, so I don't know if nVidia came around to fix the performance issues by now. I'd really love to see that.

One more thing: after porting the code over to CUDA the host code became way shorter (about 1/4 of the OpenCL code) and the whole experience from that point forward was way more pleasant...

Re:NVIDIA? please AMD (1)

Anonymous Coward | more than 3 years ago | (#35437240)

NVIDIA is dragging their heels with OpenCL. They have yet to publicly release an OpenCL 1.1 compliant driver, despite the fact they have had a beta version for about 8 months. They are also slow to respond in their forums, and many problems/bugs that were reported at least a year ago still have not been fixed. They are throwing their weight behind CUDA, plain and simple. CUDA 4.0 just came out, and has some phenomenal technologies that make me wonder if OpenCL has a fighting chance.

I think it does, but it is not going to be a short or easy battle. The fact that it is based on a standard is going to cause it to be behind CUDA for many years. However, AMD had gone full-speed ahead supporting OpenCL on both CPUs and GPUs (and is dropping Stream). Intel is developing an implementation for the CPU. I'm not sure if there is any for ARM yet or not, but that'd be another possibility.

I've written a complex code with OpenCL - a 2D Compressible Navier Stokes solver - and was able to obtain a speedup of 42X. I did not try the same code with CUDA, just due to time, but I'd love to go back and try it.

Re:NVIDIA? please AMD (0)

Anonymous Coward | more than 3 years ago | (#35436476)

OpenCL will not replace CUDA. ever.

Nvidia will officially support it but don't count on them doing any more than that. They have their own product. That product has been written by their engineers who sit next to the guys who design the chips. OpenCL will never be as fast as that sort of marriage.

NVIDIA just needs to die (1)

EzInKy (115248) | more than 3 years ago | (#35433146)

They chose to not release the necessary specs to allow others to utilize their hardware the way Intel and to a lesser extent AMD did, and as the current smartphone trend has shown, locked in is the same as being locked out.

Re:NVIDIA just needs to die (1)

gl4ss (559668) | more than 3 years ago | (#35440632)

curious to mention intel, their gpu's offer less realistic programmability.

Build your own supercomputer (0)

Anonymous Coward | more than 3 years ago | (#35434358)

I show step-by-step how you can setup your own “Super computer cluster” using Ubuntu MPI Cluster from multiple machines with the goal of bruteforcing strong encrypted passwords with John the Ripper for academic purposes.
http://www.petur.eu/blog/?p=59

And back in the day... (1)

curious.corn (167387) | more than 3 years ago | (#35435942)

... when GPGPU was in its infancy and I was lusting to play with that stuff; that's about 5 yrs ago, at most.

Alas, our semiconductor department was so content with its orthodoxy and cluster running Fortran WTF hairballs... :`(
Ah well, no point crying over that spilt milk... it just takes patience and pig headedness... :>

Hisotry repeating itself? (0)

Anonymous Coward | more than 3 years ago | (#35436214)

This sounds suspiciously like Silicon Graphics's shift from its beginning as a graphics workstation company to become a supercomputing company. It made the successful Origin line of big iron, used to own Cray, and still peddles their Altix products. Moore's Law has turned the crank a few times, but the "graphics to performance computing" change has a mighty familiar ring to it.

Re:Hisotry repeating itself? (1)

PaladinAlpha (645879) | more than 3 years ago | (#35436986)

Eh, it's a logical step. Graphics is, has been, and will always be about parallelism and matrices. Supercomputing is almost always about simulation and high-order computation, which works out to the same thing. Really good graphics hardware, thirty years ago, or now, or thirty years in the future, will always be good science hardware, and supercomputing is driven by science.

Re:Hisotry repeating itself? (0)

Anonymous Coward | more than 3 years ago | (#35438824)

I don't remember using any parallelism or matrices in LOGO [wikipedia.org] .

GTX graphics cards:64-bit floating point@25% speed (0)

Anonymous Coward | more than 3 years ago | (#35437522)

In order to ensure the NVIDIA GTX graphics card you have purchased can't be used for economical general purpose supercomputing NVIDIA has disabled 75% of its double floating point execution units.

In the GeForce family, double-precision throughput has been reduced to 25% of the full design. [nvidia.com]

Can anyone come up with a worse example of deliberate sabotage of hardware for product differentiation purposes? AMD doesn't cripple their double floating point performance.

Next to fail ... (0)

Anonymous Coward | more than 3 years ago | (#35440098)

The biggest supercomputing company is always the next one to fail ... think about it ... CDC ... Cray ... SGI ... NVIDIA next?

I missed my chance to invest in NVIDIA in 2001 when a friend of mine from SGI said, "This company is going to beat all the others in 3-D graphics because they hired the right guys out of SGI" - and I ignored him. Now it appears that if those ex-SGI employees are heading back into the supercomputing business, then perhaps my friend was wrong after all ... they hired the wrong guys out of SGI !!!

Linux support for Nvidia (1)

DMJC (682799) | more than 3 years ago | (#35440254)

Nvidia linux support is getting fixed by nouveau anyway. They reckon the GTX 5xx/4xx series is already upto the same level as the 2xx/9xxx/8xxx cards for drivers. As more resources get spent implementing opengl features in gallium and less on reverse engineering the cards, feature parity with the closed drivers will be achieved. I reckon in 1-2 years Nvidia card open source support will be at near parity with the closed source drivers.
Load More Comments
Slashdot Login

Need an Account?

Forgot your password?
or Connect with...

Don't worry, we never post anything without your permission.

Submission Text Formatting Tips

We support a small subset of HTML, namely these tags:

  • b
  • i
  • p
  • br
  • a
  • ol
  • ul
  • li
  • dl
  • dt
  • dd
  • em
  • strong
  • tt
  • blockquote
  • div
  • quote
  • ecode

"ecode" can be used for code snippets, for example:

<ecode>    while(1) { do_something(); } </ecode>