Why doesn't VRAM in SLI'ed/Crossfired cards don't stack?

Solution
Generally each card renders a full alternate frame, and thus each needs its own memory resources to draw from, the frames are then interleaved splitting the load across 2 cards.

If they were to share ram, the data rate and latency would get greatly sacrificed over the pcie bus compared to the direct connection the gpu has with it's own on-board ram, not even sure how it would work with AFR. Not only the latency but you essentially half the bandwidth with just having 2 gpu's trying to acess that. This may change in the future with dx12 where it can be shared because the gpu resources are all working on one frame together, but not sure how well that will work. It certainly wont get the same performance as each card with its own ram.
It doesn't stack because each card needs it's own ram. It's not bad for gamers, SLI is pretty meh in the first place. Buy a card with the amount of memory you need.

It may change some with DirectX12 on certain games specifically designed to deal with it and alternate SLI modes but don't count on it.
 
Generally each card renders a full alternate frame, and thus each needs its own memory resources to draw from, the frames are then interleaved splitting the load across 2 cards.

If they were to share ram, the data rate and latency would get greatly sacrificed over the pcie bus compared to the direct connection the gpu has with it's own on-board ram, not even sure how it would work with AFR. Not only the latency but you essentially half the bandwidth with just having 2 gpu's trying to acess that. This may change in the future with dx12 where it can be shared because the gpu resources are all working on one frame together, but not sure how well that will work. It certainly wont get the same performance as each card with its own ram.
 
Solution
There are details explanation about that but in general both card are working together to split thw workload evenly between them. Since both gpu can't access each other resource then each gpu need their own VRAM. Probably not the correct way to explain it but probably the easiest way to understand it like that. And while we heard VRAM stacking possiblities with low level API in reality there is not even a single demo showing how it is done.
 

The_Prince2017

Reputable
Feb 29, 2016
123
0
4,710
Only time will tell us what will happen next. Let's hope what iam2thecrowe said doesn't happen but VRAM still gets stacked somehow. *Somehow. It will be a great help for GTX 970's for 4k. 7 GB proper VRAM. If anyone goes for 970 SLI. Pascal and Polaris based cards will get rid of the issue of running out of VRAM at 4k and possibly 8k(?).

BTW: I accidentally typed that 'don't' in the title.