First post, by RaiderOfLostVoodoo
This question is going through my head for some time.
For those who don't know it:
In 3dfx SLI (Scan Line Interleave) one card calculates every even row (1+3+5+7+etc) while the other calculates the odd ones (2+4+6+8+etc).
In nVidia SLI (Scalable Link Interface) each card calculates one half of the screen. Don't remember if was top+bottom or left+right.
From my point of view 3dfx SLI is a more elegant solution, because it distributes the load evenly between both cards. In nVidia SLI, if you have an explosion in one half of the screen, one cards gets stressed while the other is idling around. That seems highly inefficient.
So why doesn't nVidia do it like 3dfx did? Are there advantages in nVidias method that aren't obvious?