Forgot your password?
typodupeerror
Graphics Hardware Technology

NVIDIA Unveils Dual-GPU Powered GeForce GTX 690 93

Posted by samzenpus
from the check-it-out dept.
MojoKid writes "Today at the GeForce LAN taking place in Shanghai, NVIDIA's CEO Jen Hsun Huang unveiled the company's upcoming dual-GPU powered, flagship graphics card, the GeForce GTX 690. The GeForce GTX 690 will feature a pair of fully-functional GK104 "Kepler" GPUs. If you recall, the GK104 is the chip powering the GeForce GTX 680, which debuted just last month. On the upcoming GeForce GTX 690, each of the GK104 GPUs will also be paired to its own 2GB of memory (4GB total) via a 256-bit interface, resulting in what is essentially GeForce GTX 680 SLI on a single card. The GPUs on the GTX 690 will be linked to each other via a PCI Express 3.0 switch from PLX, with a full 16 lanes of electrical connectivity between each GPU and the PEG slot. Previous dual-GPU powered cards from NVIDIA relied on the company's own NF200, but that chip lacks support for PCI Express 3.0, so NVIDIA opted for a third party solution this time around."
This discussion has been archived. No new comments can be posted.

NVIDIA Unveils Dual-GPU Powered GeForce GTX 690

Comments Filter:
  • The top right connector is different; any idea why this is? I also have cables that look like that, and in a moment of lazy weakness and a lack of initial comment would love it if someone cleared that up for us?

  • Sure... (Score:5, Funny)

    by froggymana (1896008) on Sunday April 29, 2012 @06:05PM (#39839817)

    But can it mine bitcoins?

  • Great (Score:2, Informative)

    by Xenkar (580240)

    It is pretty much impossible right now to get a GTX 680 unless one wants to get gouged due to the short supply.

    When will nVidia get enough chips out so my searches aren't forever out of stock?

    • by Anonymous Coward

      Use auto-notify on Newegg and if you miss it once or twice, complain. You will get the next one then. That's how I got an eVGA GeForce GTX 590 Classified last year. No need to upgrade for quite a while. It's still kickass..

    • Re: (Score:2, Insightful)

      by cpu6502 (1960974)

      nVidia knows exactly what it is are doing.

    • It isn't like they are doing this on purpose. The 680 is just a card that a lot of people want. The thing is, there's only so fast they can have them produced. TSMC is their sole supplier, and they only have one 28nm production line up and running. That line is still having some troubles (TSMC has been a bit over ambitious with its half-node plans and has had trouble at the beginning with them) so total yields aren't what they might like.

      Then the real problem is just that everyone wants a piece. TSMC has a

  • Oh man! (Score:5, Funny)

    by multiben (1916126) on Sunday April 29, 2012 @06:41PM (#39839979)
    Mine sweeper is going to look great on this thing!
  • Does anyone know if this new card will be capable of taking advantage of double precision under CUDA as is the case with some of their other high end Tesla boards?

    • by Anonymous Coward

      Yes.

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

    • by cnettel (836611) on Sunday April 29, 2012 @07:46PM (#39840419)

      They are. However, their relative FP64 performance has dropped compared to the previous generation. If I remember correctly, there is now separate silicon to do FP64, rather than just a modified path in the FP32 cores. In the previous architecture, we were down to 1/12 of FP32 performance, only a third of some of the Fermi chip cores could do FP64, and at half speed. In the new chip, the FP64 cores can do full-speed calculations, but there are only 8 such cores, versus 192 conventional cores, giving a 1/24 performance ration.

      However, Ryan Smith at Anandtech [anandtech.com] speculated that the existence of dedicated FP64 cores means that a future Fermi based on Kepler will be a mean beast, if they do a tape-out with exclusively FP64 cores. The only thing holding back double-precision then will be memory bandwidth (which would be a large enough deterrent in many cases).

      • by cnettel (836611)
        Uh, replying to myself. I of course meant that a future TESLA based on Kepler would be a beast, not Fermi.
    • Every card which supports compute architecture 1.3 (-sm13 to nvcc) or later supports ieee754 double precision, i.e. every card made for at least 3 years on a brief check of the wiki table. Your FLOPS may vary though - 2.0 is vastly better than 1.3 in this regard.
    • Only for varying degrees of "capable":
      Nvidia cripples GPGPU in Geforce GTX 680 [theinquirer.net]
      Benchmark Results: Sandra 2012 [tomshardware.com]
      NVIDIA GTX 680 Reviewed: A New Hope [brightsideofnews.com]
  • by Anonymous Coward

    As a game developer, I can tell you that the only thing that significantly affects frame rate in a GPU-bound game is GFLOPs. And as the owner of a 3-year old PC with a stock power supply, I'm most interested in the "x40" cards, because those are the highest card you can install in a machine with a stock 350W power supply.

    According to what I see on Wikipedia [wikipedia.org], NVIDIA apparently pulled a fast one this generation and re-branded some 500 series cards as the PCIe 2.0 x16 versions, while all the cards with impress

    • PCIe 3.0 is, at least in magical specifications-actually-working-as-planned land, backwards compatible with 2.0 and 1.0, though obviously only at the highest mutually available speed between the two devices.

      Is this optimistic theory a horrible pack of lies in general, are Nvidia products specifically broken in this respect, or do the newer ones make assumptions about bus speed that cause them to underperform on PCIe 2.0 boards?
      • by symbolset (646467) *
        Are you sure about that? I thought PCIe cards were only backward compatible one generation of the spec (3.0 to 2.0 for ex).
        • According to the PCI-SIG [pcisig.com](or at least their press flacks, actual standards are members only) revision 3 is compatible with both 2 and 1, and 2 is compatible with 1(excepting one minor hiccup where 2.1 increased the allowable PCIe x16 slot power draw compared to 2.0 and earlier, so there are 2.1 cards in the wild that are logically compatible with 2.0 and 1.0; but which will only function with auxiliary power).

          Internet anecdote suggests that this glorious vision may or may not actually be 100% realized, yo
  • Can someone explain to me why general purpose CPU-memory interfaces don't have this kind of bandwidth to keep the newer 6 and 8 core monsters well fed with data and code to crunch?

    • by Darinbob (1142669)

      Because gamers pay big bucks for a couple more FPS. Office workers won't get one tiny bit of speed out of a faster CPU. Scientists have real computers to use instead of PCs.

      • by Osgeld (1900440)

        "Scientists have real computers to use instead of PCs."

        really then what do they use?

        ok sure I am sure somewhere somebody has a cray ... which is powered by a fuckload of x64 cpu's, but really I bet almost all of them are using dell laptops with i7's and nvidia quadros

        • by msobkow (48369)

          That super computer/cluster market is precisely why I would have thought there would be a market for super-bandwidth CPUs. Such systems tend to use the highest of the high end processors already, along with custom memory interfaces and backbones to speed up the communications within the cluster.

          Some posters seem to have assumed I was talking about PCs. I specifically said CPU because I wasn't concerned about maintaining compatibility with desktop architectures, but the really big data crunching engines

    • There are a lot of caveats to achieving the 150+GBps theoretically available on a modern GPU, chiefly among them that all your memory read/write operations must occur in groups of 64 or 128 bytes (you can access 1 byte, but the smallest physical IO transaction is 32B, with 64/128 preferred).

      Plus, your GPU doesn't have to deal with some random manufacturer's memory chips hiding behind plug interfaces. If I take 1/3 of the ram out of one of my boxes (the furthest of 3 slots), memory timing magically tighte
      • by symbolset (646467) *
        It's not magic. The lead length to reach those far dimms is actually a prominent part of why overall memory accesses slow down.
    • by slew (2918)

      Three things:

      1. Datawidth: CPUs use one-channel 64-bit wide DIMMs (sometimes 2 if you are lucky), you can find high end GPUs with 12 to 16 32-bit channel to dram chips. Hard to find that many spare pins on a CPU package.

      2. DIMMs: People that buy CPUs want to plug in memory modules and the physics of connectors and their electrical limitations limit the performance. For example, DDR3 DIMMs need read/write "leveling" per-bit-lane compensation for clock time-of-flight across the DIMM, GPUs tend to use solde

  • TSMC's yield on 28nm has been really low. They priced it sky high because they simply don't have enough chips to make many of these monsters -- supply and demand I suppose.

    The real story in my mind is how the tech press will go gaga over a part that few will ever own and how that will inevitably help frame the entire nVidia 6xx product line and sell parts that are not the GTX 690. I guess it's no different than Chevrolet building a high performance sportscar to improve the perception of the bowtie logo.

  • ... just tell me how much it would cost for 4 of these with the SLI bridge thingie so I can make WoW run faster.
  • now I can play my xbox360 ports (that would run pretty decent with a geforce 8 series) at 180fps instead of 120, let me just shit myself

    here nvidia, have 1000 bucks!

    • It's still not going to magically fix the fact that the game is designed for a xbox360 controller and not a keyboard and mouse, though.
  • That's what GPUs are used for these days.

The flow chart is a most thoroughly oversold piece of program documentation. -- Frederick Brooks, "The Mythical Man Month"

Working...