Alienware SLI ?

ShortyShorty Manchester, UK Icrontian
edited October 2004 in Science & Tech
PcPer & HEXUS have both caught wind of an alienware SLI machine based on the Xeon (Intel) platform being available to order...
A curious forum member poked around Alienware’s site to find that they could configure an SLI system based on an Intel Xeon platform and buy it today. Shipping is still scheduled for the 1st of December, but this marks the first time we have seen SLI from an NVIDIA partner. And, we also stumbled upon a PR page that isn’t listed anywhere on Alienware’s site…

PcPer
Now, Alienware leverages a revolutionary approach to combining multiple GPUs in a single system to enable screaming, boundless performance. SLI, or Scalable Link Interface, is a high-performance technology that allows users to intelligently combine and scale graphics performance by having multiple NVIDIA GPUs in a single system. SLI works by intelligently scaling geometry and fill rate performance for two GPUs.

HEXUS

$6077 :eek: Il wait ;)

Source: [url=]HEXUS & PcPer[/url]

Comments

  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited October 2004
    This is Alienwares' proprietary SLI tho, not the same kind of SLI everyone else is using as far as I know.
  • ThraxThrax 🐌 Austin, TX Icrontian
    edited October 2004
    Correct. It is a proprietary SLI architecture unrelated to nVidia's PCIex-based x16 SLI.

    The Alienware system employs a proprietary PCI card that combines the video output from any two PCI Express cards into a complete image. It commands both cards to render a specific half of the display, so each card is doing 100% of the work on 50% of the screen. At 1600x1200, each card is rendering 800x1200. It's called "Video Array."
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited October 2004
    Please educate me. SLI?
  • ThraxThrax 🐌 Austin, TX Icrontian
    edited October 2004
    SLI! Scanline interleave. It was a technology that 3DFX developed back in the mid nineties with the advent of the VooDoo2 series of graphic accelerators. The basic gist was that every card was capable of rendering alternating lines of an output provided another identical card was present.

    Each VooDoo2 came with 1 female and 1 male CRT port on the back of the card, each card also came with a male to female VGA cable. You would plug the male end of one of the VGA cables into one of the cards, and the female end into the other card. Then you would take the second cable and connect the second card to a 2D adapter (Preferably a Matrox Millenium 2). You would take the remaining open VGA port on the first VooDoo2 and connect your monitor to it.

    The 2D accelerator (A required device with or without SLI) would handle 2D imagery, while the twin VooDoo2 cards would render every other line of the display. If you were in 640x480, each card was only rendering 320x240, thus increasing your performance by (literally) 50% or more. Having two VooDoo2s also unlocked 1024x768 acceleration for many games and applications.

    Present-day SLI, as designed by nVidia, was heavily influenced by the 3DFX engineers that remained within nVidia Corp. after nVidia bought them out in the late nineties. Today's SLI is achieved via integrated architecture that allows both cards to negotiate with one another provided you've connected them via a tiny adapter that connects both cards on tiny slot interfaces opposite their PCIEx connectors. It is unsaid how the SLI is achieved, be it alternating pixels, screen halves, or scanlines; what we do know is that it's hardware based, and seems to require identical cards. The cards also load-balance so neither GPU is doing more than it should.
  • CyrixInsteadCyrixInstead Stoke-on-Trent, England Icrontian
    edited October 2004
    I saw somewhere that nVidia's new SLI technology worked by each card drawing half the screen. The problem they had to overcome was synching the two so that there was no tearing and they were in sync.

    ~Cyrix
  • ThraxThrax 🐌 Austin, TX Icrontian
    edited October 2004
    I think I recall that too. I know the technology's rendering style was different from 3DFX's, so I know it wasn't alternating lines of rendering. The only other design that makes sense is halves of the screen, so you're probably right. :)
  • SputnikSputnik Worcester, MA
    edited October 2004
    they were trying to optimize it such that each card was doing the same amount of work.... the example they gave was in a far cry style environment with lots of pretty vegitation and land on the lower half of the screen and a nice boring, easy to render sky up above. the first card would do the top 3/4s of the picture (as the sky takes virtually no resources to render) and the second card would render the lowest 1/4 of the screen, thus sharing the load. basically network load ballancing but with video.
  • MedlockMedlock Miramar, Florida Member
    edited October 2004
    I forget where i heard it now, but each card is responsible for half the screen, just about. Actually, one half of the screen could be more complex than the other, so if, for instance, 2/3 of the screen is of the same complexity as the remaining third, then that is what each card will render. This technology equally balances the load, not the exact proportions of the screen.

    EDIT: sputnik posted while I was typing... I didn't see that. I do remember that example, though...
  • CammanCamman NEW! England Icrontian
    edited October 2004
    Alienware sucks...but SLI is cool. And one correction, nvidia didn't aquire 3dfx in the 'late nineties' , it was 2000 or 2001 when that occured, after the Voodoo4/5 cards were released.
  • SputnikSputnik Worcester, MA
    edited October 2004
    Early December 2000 to be exact. Got a card that christmas, went to get drivers and got a nice press release: We dont exist anymore, click here for drivers!
Sign In or Register to comment.