• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Triple SLI Titan's not enough for 4k surround

Actually, you'd almost need 4GB with everything maxed in Crysis 3 on a single 4K screen.

That is what I am saying.:)

There is no cards now and probably not even when Maxwell is with us that will be able to run three x 4K screens. Even if Maxwell is 50% faster than the Titans/290Xs it still won't be enough GPU grunt even using 4 top Maxwell cards.

Some of the guys on these forums are using 4.5gb of VRAM on their Titans @5780 x 1080 which is a lower resolution than 4K. With the next gen of games VRAM is also going to be a problem.
 
4k surround is terribly silly. Use 17 million pixels worth of display clarity and GPU processing for blurry peripheral vision...
 
Isn't Mantle supposed to stack the VRAM? So 2x290's = 8GB - 3x290's = 12GB or have I misread this?

It allows developers to implement their own multi-GPU configs, but AFAIK it's unlikely we'll see stacked VRAM anytime soon. Johan mentioned using 1 card for rendering and others for compute tasks, AI, etc. As opposed to just AFR which we have now. If I've understood it right we could see >2 GPUs become much more prevalent as you could maybe have 2 in AFR and another for purely logic, not drawing anything but accelerating portions of code that are well-suited for a highly-parallel system like a GPU. We have already seen something akin to this with hybrid PhysX, where an ATI crossfire setup could run alongside an Nvidia card.
 
Last edited:
I don't know that much about mantle but to do what you describe above would change the way CF works.

I just had a look about and can't find anything, so assume me as wrong on that :D

It allows developers to implement their own multi-GPU configs, but AFAIK it's unlikely we'll see stacked VRAM anytime soon. Johan mentioned using 1 card for rendering and others for compute tasks, AI, etc. As opposed to just AFR which we have now. If I've understood it right we could see >2 GPUs become much more prevalent as you could maybe have 2 in AFR and another for purely logic, not drawing anything but accelerating portions of code that are well-suited for a highly-parallel system like a GPU. We have already seen something akin to this with hybrid PhysX, where an ATI crossfire setup could run alongside an Nvidia card.

Cheers Tangy. I skipped through a couple of the readings on Mantle and came up with 5. :)
 
Just thinking about the latency involved in accessing another gpu's vram over pcie... you could potentially do it but i cant see that you would ever want to without replacing pcie with something substantially faster

i vaguely remember reading the same thing somewhere but having thought about it i dont think its a great idea, it might be better than running out of vram altogether, but it will still be a hit on performance
 
The problem is fillrate...... There's just not enough of it!

That's why the 290 % 290x beat the Nvidia cards by a decent margin at 4k because they have 25% more fillrate.
 
Am i being blonde, three titans running three screens equals 25-30fps, so one titan running one screen would be 25-30?

Edit ignore me just reread chapparels post i'm being blonde, now where did i put that dye :D
 
If is capable to do so, the R10 series going to fly of the shelves, if there are any Mantle games out :D
 
Back
Top Bottom