Alienware Video Array Technology Interview


posted 6/7/2004 by The GN Staff
other articles by The GN Staff
One Page Platforms:
GamingNexus: Does it mean that the one portion will get smaller or will more cycles be sent through the card that is under load?

Brian Joyce: It’s actually by proportion. For example, more first person shooters are a little more graphic intensive on the bottom portion of the screen because there are more enemies and things going on down there. So, the card drawing the top half has less to do than the card drawing the bottom half. So, if that’s the case, we would increase the portion that the top video card was drawing to balance out the load.

GamingNexus: So instead of the rendering line being at the halfway point would actually shift down?

Brian Joyce: Exactly. So, instead of being 50-50, it becomes 60-40 or 70-30, etc. It happens on a frame by frame basis. Since video games are essentially like a movie where you have multiple frames per second, the fact that we are doing this by frame means that your eye will never see the adjustment.

GamingNexus: This is kind of a stupid question but can you use different speed cards or do both cards have to be exactly the same?

Brian Joyce: The answer to that is that they would have to be from the same manufacturer since the drivers would have to be the same. It is ideal to have them paired, i.e. the same card. Theoretically, you could use two different cards but one of them would be constantly making up for the other one because one of them is faster than the other. So, if the image was equally loaded from top to bottom, the fastest card would end up drawing more of the screen than the slowest card.

GamingNexus: Your announcement and demonstration of the Video Array was a pretty big announcement at E3. How has the community responded to the technology? What about from the major game developers?

Brian Joyce: The community and the press that saw it were extremely impressed. There’s a sense of disbelief from those who didn’t see it because the concept has been attempted before, not in this particular execution, but the multiple video card/one image thing has been attempted before but not executed very well. From major game developers, especially those who saw it at E3, they are extremely excited about it and want to know how soon they can get their hands on one so they can start coding to take advantage of that capability.

GamingNexus: The technology seems a little similar to 3DFx’s old SLI configuration except for rendering every other line, you do the rendering in halves. How does that old 3dfx technology compare to your technique?

Brian Joyce: SLI stood for Scan Line interface where each card drew every other line of the frame and my understanding was that the major challenge was to keep the image in sync. If one line’s longer than another, then tearing, artifacts, and keeping the two cards in sync was a real issue. The benefits of doing it half and half is we can take advantage of the load balancing and the synchronization challenge can be overcome.
Page 2 of 5