Not dissing Vega or anything but I err on the side of caution with his build threads. He had previously released an article complaining about the bandwidth definiciency of AMD's crossfire finger solution acting as a x1 lane for througput for communication, and came up with "fancy" charts to support his data saying how Nvidias mutli connection sli fingers are better than the single fingers for crossfire, in quad and tri crossfire scenarios. I used his exact chart 5 months back when I defended Nvidia for their choice of multi connections.
Another extreme user with a quad 6970 and multi 1600p screen setup proved me wrong... oh so wrong... and said Vega is full of it.
Since then I don't trust Vega as much.
I say you wait until the launch of the 690 or 7990 and see what users think. Don't trust JUST Vega (I'm sure you aren't just trusting him), AFAIK he has no affiliation to any tech site, and even as an unbiased 3rd party source, his data could be questionable. He is a user with a lot of knowledge and money for the expense for high end setup. Right now he is setting up a fresnel lens setup just for the FW900 in Surround/Eyefinity. The man knows his stuff, I don't doubt that, but his data may not always be correct.
That's not saying that all technology sites are correct on the other hand. comparing the 5850 vs 460 sli two years back saw clashing data between HardOCP vs such sites as TPU or Anand. Turns out Hard's drivers were older and did not match data posted by other sites on the matter, and they later rectified the issue August of 2010 (If I recall).
Take his data with a heaping pile of salt. We are going by theory here that the bandwidth demanded by a single 680 will surpass demand by pci-e 2.0 x8. And the way data transfer is done (as seen in the wiki article) from 2.0 to 3.0 is different, it is after all an evolution.
New features for the PCIe 3.0 specification include a number of optimizations for enhanced signaling and data integrity, including transmitter and receiver equalization, PLL improvements, clock data recovery, and channel enhancements for currently supported topologies.
Originally Posted by PCI-SIG FAQ
PCIe 2.x 5.0GT/s 4Gbps ~500MB/s ~16GB/s
PCIe 3.0 8.0GT/s 8Gbps ~1GB/s ~32GB/s
...PCIe 3.0 is the next evolution of the ubiquitous and general-purpose PCI Express I/O standard. At 8GT/s bit rate, the interconnect performance bandwidth is doubled over PCIe 2.0, while preserving compatibility with software and mechanical interfaces. The key requirement for evolving
the PCIe architecture is to continue to provide performance scaling consistent with bandwidth demand from leading applications...
I would still say wait, and go for an mATX setup. If when it releases (690 and I presume the Maximus V or the Sniper M2) and it is shown with sites that you do indeed need a plx replicator chip, you might be short changed and I feel sorry for you as I do like mATX setups.To clarify
: I don't say for one second I am right
. I could very well be extremely wrong
and full of crap.
Often times I am. My goal here is to help you justify purchasing these gpus come release and to make sure they are utilized. I am 100% on your side BHK, so don't take my posts in any offensive way, for I did not intend for them to be such. I'm here to learn myself and to help you get your questions answered. If you post a piece of information that goes contrary to all this hypothetical bs I'm pulling right now, and proves that indeed 680's are crushed by the lack of throughput by x8 2.0 then I will be the first one to agree with you and change my views.
BAM: OCN THREAD
questioning Vega. That discusses bandwidth, not so much in the amount of transfer from a 680's compute, but bus limitations and memory demand of a mutli screen setup.
Vega's threads in question:
odd... I can't seem to find the EVGA thread from Jan '11 by Vega anymore
In the end of the day:You said you had more links. Could you please post them. I want to be clear about this whole bandwidth debacle, that way I can give people proper and not misguided advice for multi gpu setups. This is to clarify any personal ignorance of mine. OP if you could do that, I would be grateful. Thanks.
Man, I spent a good twenty minutes looking for that OCN thread from Levesque, and another 15 hunting for Vega's old threads. 35 minutes spent looking for links. I should spend more time on other things lol.
Edit once more:
Interesting.. simply googling PLX brings up quite a bit of information. Not all related to your question though, but very informative. Tell me OP, have you been into computers long, or were you just always inquisitive. When I first started not too long ago, I was ignorant as hell. Most people were. I didn't know (and still don't know) even the top layers of computer technology. I'm surprised seeing a new members asking such questions. Do you frequent other forums and you just recently joined OCN? I'm impressed by certain new members who have knowledge, often times they having prior experience with this technology. Such users make me not just classify new users as total dolts, but think that yes the general majority is ignorant, but there are others who are not.Links about PLX technology (Click to show)
Edited by Spykerv - 4/12/12 at 1:24am