Latest Quadro FX graphics card has 4GB of RAM

The last time Nvidia shipped a product called “FX 5800,” things didn’t turn out so well. This time, Nvidia has brought back the FX 5800 moniker to define its latest and greatest Quadro professional graphics card, which packs a whopping 4GB of memory and seems to be based on the same GT200 GPU as the company’s fastest desktop offerings.

The Quadro FX 5800 is launching at $3,499, and its spec sheet looks well furnished: 240 stream processors, fill rate of 52 billion texels per second, 128-bit precision, true 10-bit color support, 102GB/s memory bandwidth (which should translate to an 800MHz GDDR3 clock speed, if my math is right), and “interactive 4D modeling with time lapse capabilities.” Nvidia says the Quadro FX 5800 should be well suited to applications like “oil and gas exploration, medical imaging, styling and design, and scientific visualization.”

For reference, the GeForce GTX 280—Nvidia’s speediest desktop card—has a 48.2 billion texels/second fill rate and a gig of RAM clocked at 1107MHz. Nvidia’s numbers suggest the Quadro FX 5800 might actually be more power-efficient than its desktop counterpart, too. The company quotes maximum power consumption of 189W for the Quadro and “Maximum Graphics Card Power” of 236W for the GTX 280.

Look for the Quadro FX 5800 in future workstations and from Nvidia’s usual professional GPU partners, including PNY Technologies, Leadtek, and Elsa.

Comments closed
    • kilkennycat
    • 12 years ago

    Recession… recession…. what recession?

    These products are virtually recession-proof and have a huge profit-margin for nVidia. Advanced R&D always goes on even in a recession, and it is to those high-tech industries and academic institutions than this product-family are targeted. The double-precision data- paths/computation-cores in the GT200 ( the first with DP in any GPU family), together with the massively-parallel compute-acceleration offered to applications developers by the CUDA toolset, make these products an essential element in advanced scientific or engineering research, regardless of whether they are used for graphics imaging or not. Having powerful desktop computation available instead of waiting in line on a central server can be a huge time-saver.

      • UberGerbil
      • 12 years ago

      This is in response to what, exactly?

      Companies that want something like this for computation, as opposed to visualization, will just buy Tesla.

      • indeego
      • 12 years ago

      I wonder if they are depression proof? Three friends in different industries lost their jobs or had hours scaled way back on Fridayg{<.<}g

    • Vasilyfav
    • 12 years ago

    Buuuut…can it run Crysis?

      • eitje
      • 12 years ago

      Buuuut…can it fold?

        • Peldor
        • 12 years ago

        Not only fold, but spindle and mutilate too.

          • shaq_mobile
          • 12 years ago

          buuuut… can it give me a happy ending?

            • stmok
            • 12 years ago

            That’ll require a driver update.

    • UberGerbil
    • 12 years ago

    I haven’t had any reason to look at pro cards in quite a few years now, but back in the day the OpenGL drivers for them didn’t include any of the usual “approximations/optimizations” that the gaming-oriented drivers featured. That meant they ran slower, but they were more accurate. I remember seeing screen shots of the same complex hull rendered on both the consumer and pro drivers, and you could see the gaps between triangles on the one rendered with the consumer drivers. That’s not going to matter for gaming and not even for video work, but if you’re doing architectural/engineering work you definitely don’t want any corners cut.

    How true that still is I don’t know, but I wouldn’t want to be the design firm on the hook for extra time and materials because something didn’t fit together in real life that looked like it would on the model.

    §[<http://www.wired.com/wired/archive/12.11/gehry.html<]§

    • Ryu Connor
    • 12 years ago

    This card clearly needs a 32bit OS.

      • 5150
      • 12 years ago

      Thread over man, thread over.

        • shaq_mobile
        • 12 years ago

        where is bill paxton when you need him?

      • UberGerbil
      • 12 years ago

      Sure. They can run it on the card.

      • Kurotetsu
      • 12 years ago

      I wish I could see this as just a joke….

      I’ll go so far as to say you could bet money and have pretty good odds that some ‘enthusiast’ will buy this card, throw it on his XP Pro/Vista 32-bit machine, and then run screaming into this or some other tech site complaining about how Windows keeps crashing on start up.

    • PRIME1
    • 12 years ago

    Oh how I’d love to work in an industry that actually uses gear like this.

      • no51
      • 12 years ago

      That part is easy; finding a company that would buy you gear like this is hard.

        • PRIME1
        • 12 years ago

        These days you’d be lucky to find a company that will pony of up for integrated video from Intel.

          • moose17145
          • 12 years ago

          Sigh…. sad but true

    • zqw
    • 12 years ago

    Generally, tested/certified drivers for pro apps, and very good support. Both of which can be worth the extra cost at a big shop. There are also a few niche hardware features that get enabled such as 2-sided lighting, line AA, and hardware overlays. In the case of Maya (which I use), these features have long been replaced so “normal” graphics cards (like in laptops) work well.

    • cegras
    • 12 years ago

    I’ve always wondered what is the exact difference between these cards and their consumer counterparts?

      • MattMojo
      • 12 years ago

      Certified CAD drivers or something like that I suppose…. I am shooting from the hip here though.

      • GFC
      • 12 years ago

      Drivers. They use sh**load of time in doing all the professional drivers, the cards don’t realy differ actualy.

      • moriz
      • 12 years ago

      they are basically identical, except for a line of code in the vBIOS. this is why it’s so easy to BIOS flash a quadro to a geforce and vice versa.

        • Umbongo
        • 12 years ago

        It isn’t easy to flash them (once someone has worked out how it is obviously) and when you do you do not get full functionality, you get the ability to run quadro drivers on a geforce and get some improvement on some applications (usually not openGL ones).

          • Kurotetsu
          • 12 years ago

          I’ll just add to the echo here…

          From what I’ve understood, for the massive premium over the standard desktop model you application-specific driver profiles and onsite tech support (meaning if something goes wrong, either with the card or the drivers, a tech will actually drive/fly over to your place of business and fix it for you).

          You basically get the kind of driver development (as in, they release a finished driver for their Quadros while throwing out beta-level crap for everyone else) and support you really should be getting as a standard.

      • Joerdgs
      • 12 years ago

      The ridiculous price.

        • Meadows
        • 12 years ago

        Keep in mind that these cards come with the most memory ever (at a given time at least, technology progresses), not even that Palit 9400 GT with 1 GiB would do what this one can. Memory can be essential to a lot of production/creative workloads and since this is stuff for companies, they’ll be able to pay for the speed+assload-of-RAM combo anyway.

      • stmok
      • 12 years ago

      You are basically paying for a guarantee that this hardware and its driver will work with specific professional applications. (In addition to tech support).

Pin It on Pinterest

Share This