Nvidia CUDA Capable Graphics Cards &
Quadro High-Efficiency Professional Video Cards

NVIDIA Quadro® cards support NVIDIA CUDA™ technology. Although the newer NVIDIA GeForce® cards also support CUDA, the GeForce cards tend to focus more on the desires of gaming users (speed at any power consumption / Wattage level) whereas the Nvidia Quadro line delivers outstanding performance at substantially lower energy levels, making the Quadro cards much more energy efficient than the GeForece cards for performing CUDA operations and other workstation graphics operations. In fact, some of the Quadra cards are so efficient they are fanless (i.e., passive cooling alone is enough).

In addition to power efficiency and CUDA support, various Quadro cards offer the lastest DisplayPort interface in combination with multiple monitor / multi-display capabilities for up to 4 high-resolution simultaneous displays.

We here at "Modern Visions for Technology" (MV4T) realize the incredible potential that NVIDIA Quadro® Professional Graphics Cards offer our workstation solutions; they provide a perfect complement to our high-performance and high-efficiency CPUs, Power-Supplies, Storage Devices, and RAM -- all yielding totally Green and Powerful workstation solutions. The Quadro cards we use deliver great performance and low power consumption (Watts).

NVIDIA Quadro® professional solutions deliver the fastest application performance, the highest quality workstation graphics and are fully certified on all industry-leading professional applications. Beyond just raw performance and quality, NVIDIA Quadro takes the leading computer-aided design (CAD), digital content creation (DCC), and visualization applications to a new level of interactivity by enabling unprecedented capabilities in performance, programmability, and precision.

We have assembled the following specifications table, comparing CUDA graphics cards, from information we could obtain publicly from various sources like NVIDIA's web site, various press releases, and technology reviews. Although we recommend the NVidia Quadro products, since they fit within our energy-efficient computing and computer hardware focus, we have also added information about the CUDA-enabled GeForce products so we could offer a comparison between Quadro and GeForce graphics cards from a CUDA and power-efficiency perspective too.

NOTES: In the following table, these conventions are used:

• Product: refers to the Nvidia's product name for graphics card, and not just the GPU. If the card is a MULTI-GPU type (as indicated in parens after product name), the CUDA-Cores are PER-GPU, as are the memory and bandwidth values.
• Type: (C/E) Card and/or Embedded-Only solution. Embedded implies that you would have to purchase the Product as part of another product (e.g., quite a few of the Mobile CUDA graphics solutions are embedded).
• Cores: refers to the number of "CUDA cores" or "Stream Processors" - each being equivalent.
• Watts: refers to rated maximum wattage the product draws. Also, if specified, the suffix -P and/or -A refer to whether the product employs Passive (i.e., fanless) and/or Active cooling. Default is likely to be Active (i.e., fan cooled) if neither specified. A question-mark (?) following the Wattage figure indicates best-guess information, compiled from online sources other than Nvidia.
• Bus: the interface type; e.g., PCI-Express, PCI Express x16 (aka PCIe x16), etc.
• Memory: column shows max RAM product can accommodate, and the memory-bus width, and graphics memory bandwidth
• Performance: the rated raw Processing for Texels per Second/Fill Rate (in Billions), or the raw processing power in GigaFlops (GF*) if Texel Fill rate not available; this is followed by the rated Triangles per second drawing speed if known (in Millons).
• Display Interfaces (and maximum screen resolutions): e.g., DisplayPort, DVI, HDMI, HDTV, VGA,... are typical interface types; Maximum digital Display (D) resolution and/or maximum resolution for Analog (A) display support (both are in horizontal-pixels by vertical-pixels per screen) - if neither an A or D is shown, it is likely that Digital is the highest resolution option. In general, these graphics boards support multiple independent simultaneous displays, with some models supporting 4 displays or more displays. A leading number in parenthesis before a resolution value indicates the card can achieve this resolution on that number of displays simultaneously.
• Standards: OpenGL version, Shader Model, DirectX