mobhill
...always seemed to be problems.
...I'm sure it could have worked out if it was focused on more...
...but truth is developers aren't going to spend a lot of time developing for the 0.01% of people who would really use it.
I wonder why only a microminority of people have been using it? It works so well!
You don't think that, if SLI--which is
not an especially difficult problem--had been properly implemented, and a person with a mid- to top-range card could have had an instant ~60 to 95% increase in framerate by simply purchasing a cheaper second card, new or used, that there would have been a LOT more people taking advantage of it?
Of
course there would have been.
This is a cart-before-the-horse viewpoint. Gamers can't
use anything before it's
available to use. So taking an option that has had disgusting support for well over ten years, and claiming that the reason it's not better is that no one uses it is entirely nonsensical. Of
course no one's been using it.
(My first and last experiment in SLI--two GTX 260s--got me a 23.3% framerate increase at 1920x1200 in the game for which I needed it the most:
Crysis. None of the other games for which I was below 60 fps on a single GTX 260 had great scaling either.)
As you correctly stated, it
could have worked just fine. The greatest blow to SLI was the Great Consolization of 2008, in which AAA PC development effectively ended, and all major AAA games were henceforth conceived, designed, and developed for console hardware, and the sensibilities of the console gamer. Among the very last true AAA PC releases were
The Witcher: Enhanced Edition and
Crysis, both in 2007.
But even before that (as evidenced by the lack of proper support in Crysis which I mentioned), proper industry effort and coordination to maintain a solid protocol and straightforward implementation had not occurred. Though I wasn't a computer/PC gamer at the time, my understanding is that SLI, whether "Scan Line Interleaving" or other methods, was originally quite solid and functional. As video card hardware changed, and rendering became more complex in the '00s, it was not prioritized by an industry that was increasingly consolidated into a few huge developers, and increasingly interested in console-based Lowest Common Denominator development in order to maximize profits at the expense of innovation and creativity.
But again: If SLI actually
worked, it wouldn't just be 0.01% of people who would really use it.