It's nice to dream.

[mrSHEiK124]

nVIDIA's most powerful single slot card.

Here: http://www.amazon.com/gp/product/B00...FRY7ZFSM5BWRBA

7 of these. Two Xeons. SR-2. That would be fun.

 

[theAnimal]

Or just build a quad 12 core (16 core when BD finally arrives) AMD for higher PPD & lower cost.

 

[Sunny129]

what exactly is SR-2?

 

[mrSHEiK124]

what exactly is SR-2?

eVGA dual Xeon board with 7 PCI-E slots.

 

[LOL_Wut_Axel]

nVIDIA's most powerful single slot card.

Here: http://www.amazon.com/gp/product/B00...FRY7ZFSM5BWRBA

7 of these. Two Xeons. SR-2. That would be fun.

Eh, not really. You'd need at least two power supplies for it and you have to allocate 2 threads or so to those seven GPUs. Also, the single-slot KFA2 GTX 460 has more CUDA cores. Don't know if the Quadro cards have an advantage at folding, but I doubt it.

 

[Sunny129]

eVGA dual Xeon board with 7 PCI-E slots.

ah, thanks for the explanation.

Don't know if the Quadro cards have an advantage at folding, but I doubt it.

i think the advantage would come into play with projects that require double precision calculations (not sure if Folding requires such a GPU). and even then i'm not sure if the Quadro cards are any better than the GeForce cards. now the Tesla cards on the other hand DO have better double precision performance than their GeForce counterparts.

 

[Ken g6]

Funny, I was just calculating that four of these cards on a more standard board, in a case with extra space, would provide 4096 cores!

 

[mrSHEiK124]

Funny, I was just calculating that four of these cards on a more standard board, in a case with extra space, would provide 4096 cores!

that's 80-100k PPD. Who folds for points though?

 

[Lightflash]

I would rather get a Quad G34 and save my money. And as far as I know the Quadros perform on par with regular GTX xxx for way more price for extra memory and other features.

 

[Sunny129]

I would rather get a Quad G34 and save my money. And as far as I know the Quadros perform on par with regular GTX xxx for way more price for extra memory and other features.

this is what i was afraid of. i'm also skeptical that the nVidia Tesla cards would be any better with regard to single precision performance...but they would for sure be the way to go for DC projects requiring double precision GPUs b/c their FP64 performance is 1/2 that of their FP32 performance (unlike the GeForce GTX GPUs, whose FP64 performance is limited to only 1/8, and sometimes 1/12, of their FP32 performance).