Dual 560 GF 114 card soon ?

[scruff]

This could be interesting.KFA2 plans dual GTX 560 Ti graphics card | bit-tech.net

[_dangtx_]

smart. do a trial on dual 460's, then drop 560 cores in there. sure itll be a wee bit longer, and AFTER a competitor has launched dual 460 card, but dual
560 will be much juicier.

[draemn]

Yep, but I've heard that part of the reason to stay with dual GF104 over 114 is because there is still a lot of stock of the 104 chips.

[Arinoth]

Would help me avoid the limitations of sli if this is a dual 560 that can perform better at high memory bandwidth tasks

[scruff]

Originally Posted by Arinoth

Would help me avoid the limitations of sli if this is a dual 560 that can perform better at high memory bandwidth tasks

Not trying to be a smartypants here but what do you mean by SLI limitations , I always thought that it scaled better than crossfire although I would also never expect double the performance running 2 cards from either camp . Everything that I have read says that , at least for me anyway, SLI or Crossfire is worthwile ( as in I get a big enough boost with 2 cards that I'm happy)

[_dangtx_]

i believe 6 series crossfire does a bit better than 5 series sli.

[Arinoth]

To my knowledge and what has been explained to me, in SLI each card still has some of its features that are run separately. One of those is memory, they do not share it between the two so that anything that has high memory usage or bandwidth is still limited to 1GB per card, whereas a dual card has that entire 2GB memory shared across the 2 gpu units, lowering memory bandwidth usage and increase performance

[SKYMTL]

I'll expand a bit on what Arin said.

Basically, dual GPU technologies don't allow for a fundamental "doubling up" of performance in terms of memory utilization. AFR or Alternate Frame Rendering allows GPUs to work together to render a 3D environment but each GPU will render SEPARATE frames. In a dual GPU config, this means GPU #1 will render frames 1, 3, 5, etc while GPU #2 will render frames 2, 4, 6, etc. this is also why sometimes you see "flickering" as a driver error makes one GPU skip certain frames. This means even though each GPU has say 1GB of onboard memory, the total accessible memory at any one time is NOT 2GB. Rather, each GPU will have access to its own memory so the same bandwidth constraints will be in place but rendering times should (in theory) be cut in half or thereabouts.

Naturally, load balancing and partial resource pooling at the driver level allow for increased overall memory performance in every scenario but usually the same bandwidth limitations will be in place.

[scruff]

Ok thats basically the way that I understood it as well . As I said earlier that I wouldn't expect 2 cards to run twice as fast as 1 because there has to be some overhead penalties when you put 2 cards together . Thanks guys