I will be getting my 3rd replacement card this monday, The last card i RMA'd had the exact same problem you describe But the first card I RMA'd did not.   So common sense is telling me there is just a problem with these cards. I am trying to run a SLI setup and do have one working gtx 460.
 
When i sent in my cards for RMA they also told me the card passed testing. Some people are forced to send in their Entire Rigs to the card manafucturer just so they can then tell them the videocard is at fault. But you can take that same card that does not work in your system and put it in one of their older systems and would run fine.
 
Evga Claims to not be aware of any such issue, But Nvidia IS and is investigating the matter right now.  So more then likely its just a problem with the 400/500 series cards.
 
 
 
 
 
 
 

Thanks a ton morphaus for that link.
 
Every time I trick my 480 into working it dies on me a month later. I can either spend $300 on an HD6950 (With less performance) or $200 on a new motherboard that works with the 480. Damn this 480 is getting more expensive by the day.
 
 

Morphaus all I see you doing on this forum is bad mouthing evga in every thread (ASK YOUR SELF WHY).  Fermi might be power hungry, (WHAT?) and run hot but your personal troubles are not indicative of every fermi owner.( Good to hear, Not ever one haivng these's problems.
But after months, of trouble shooting... ITS the Cards. with certain motherboards.( my Case)
 
Well i did  spend £1100 .And if they work solid like 280 sli, i would not behere typing, Have i got anything against EVGA, well yes Ive used RMA, corsair. dell benq, samsung, phillips.ECT  And this problem is the cards EVGA and this Motherboard EVGA. RMA is poor in euro.. Delt with alot in pass 10 years.    From Differant company's  (plus ive got to spend £ to send to germany. Imagine the cost, motherboard and 2 cards....... All evga brand. 
 
I did tell EVGA about my problem. But No reply ish. Did get email about Change for (standard. 580's)
But that no Upgrade at all, From 480 FTW SLI Watercooled cards. if i got the new cards, after shipping cost's (charge me) and to buy 2 extra blocks, cards turn into not £1100 but £1300... how would you feel.... 
 
 

Morpheus did help me out, so I think he deserves a little credit ;) He didn't bash EVGA in the process, either. 
  
HRPuffnstuff: any thoughts on my issue or did you just drop by to harp on Morpheus?  
 
well ive had 16 weeks with this issuie, All i ugraded was psu and gtx480sli. thought it might have been drivers, Ive done everthing i can at my end. Me Im bashing, THE RMA With evga in euro, why? sent RMA in. got email saying explain prob, Sent http://forums.nvidia.com/index.php?showtopic=187401&st=0 and others links. no reply from RMA
But did get reply to step up too 580s sli, but for me its a step down Why. cost to send too germany. +(down time) cost to buy 2 new W?C blocks. All my, RMA's,  ive never paid at my end.. and never will. if the problem is from who i got it from.... And ive got EVGA M/B and cards.
 
since 2001 add 10 + RMA's ( differant company's) never had to foot the bill. and to show this propblem i would need to send M/B and 2 GTx 480 in post. Cost, My end alot+ downtime.
 
Could pop in 280s and send back 480's.... But we carnt, Why  http://forums.nvidia.com/index.php?showtopic=187401&st=0     
 

Yeah sending back 2 cards, 400 series has not helped many.they still got the problem. Nvidia is working on the problem. But they carn't reproduce the problem. Yeah right. Yeah its nice to not be alone with this problem, after paying £1100 2 cards.  Yeah if i step up. Might have  find extra £200 ish. shipping and 2 new water-cooling blocks. + downtime.
 
Been thinking about sandybridge. ? idunno......   
 

Sad these card manufacturers seem to be giving us the run around. When i talk to tech support it seems they are reading lines out of a book making it even hard to get a word in sometimes because they talk without taking  a pause, A typical telemarketing tactic.  I find it odd Tech support would incorporate such a tactic.

Obviously there are lots of blame to throw around, some lies with nvidia and some with the user.
 
I can tell you in most systems a single gtx480 is not a problem, but add the heat and powerdraw of the SLI setup and most endusers cannot engineer out the solutions needed. 
 
Just face it, some hardware is incompatible. Get a board you know works with it and get over it instead of the same stuff over and over. The same applies to Cases, PSUs, fans, etc.

Nvidia has alot of failings in the FERMI release and in driver issues etc. Dont expect them to fix anything. If it doenst work and you tried move on or get left behind. I wouldnt spend more than a week trying to make something work before admitting that i dont have the knowledge to do it. You arent Nvidia engineers and so fort. RMA, buy something else, or switch brands.
 

Thanks donta. I did try and address those issues. I've tried 3 PSU's (Two of the same brand / wattage, and one new one) and actively monitored voltages and temps. I've done everything but re-seat the motherboard. Since removing the card, the system's been stable. 
 
Between that and seeing so many others exhibiting the same issue, I feel confident that it's a problem with the card.
 
For those who feel so inclined to personally defend Fermi / nVidia, I'm not saying it's a bad card: I'm pointing out a problem that needs to be resolved. If the card was "bad," then so many users wouldn't be using the card successfully. If the card was "perfect," though, so many users wouldn't be exhibiting the same problem. I think we can agree that the card may have some issues that need to be resolved, even if they don't affect everyone.
 
On a personal note, I don't think that something this widespread could have gone unnoticed by nVidia for so long. I am disappointed in how nVidia is handling this.

Look Everyone of you Should have Known what FERMI GTX4XX was from the start. A Failure. It failed to enter the market when it was supposed to, was not at all a 512 core design. The yield of functioning chips was <3% from the silicon at the foundry.  And those 3% represented the unfinished 512 fermi which was respun  once if not twice more and entered the market finally in late 2010 versus Q3 2009. 
 
The 400 series is and always was a stopgap between the 200 series and the 500 series, and we wont even talk about the rebadging they did trying to come up with a 300 series.
 
Every Review and Preview predicted it would be too hot, it would be a power hog, and NVIDIA would likely LOSE money on it. But They were so late to the Game it had to be Released.  ATI 58XX cards already supported DX11 and tessellation. So gtx4xx got rushed out a full 6 months or more before the True Fermi (gtx 580)  was released. 
 
Yes each of us here in this Forum purchased an engineering failure. Dont expect much in the way of sympathy from Nvidia.
 
These chips are all Binned based on capability. It physically as produced could not run the full 512 cores because of HEAT and Power requirements. So they Neutered a group of 32 cores off of each piece of silicon, Meaning the 480 has 15 SMs versus the Fermi Spec 16 SMs.
If it ran within spec with 480 cores of the 512 it got to be a gtx480 so maybe 5% met that criteria, If it ran with less then it was maybe another 3-5% That met gtx470 spec, and then 3-5% that may meet gtx465 specs and on and on down.
 
This is why alot of chips vary greatly among voltage and thermal properties even among the same core class. The manufacturers then take and test these chips and the ones that spec higher with greater clock speeds without needing extra voltage of course become your SSC, SC, and FTW chips.  but even some of these chips will experience problems because of the HEAT the chipset produces.  Heat makes for the need for added voltage to stabilize and voltage makes heat causing instability in a vicious cycle that becomes hard to balance when running at temps far greater than prior chip designs and specifically within a Flawed design. 
 
 

Heres a thought , RMA your MB.
Obviously if you had 6 working months and then something fried. Do not limit your damage strictly to that component. Possibly something fried in the PCI bus that or MB power that will still allow enough for the old card but not for the new cards. If something happened to the PC say a surge and it was damaging to the VGA card then you can and will have other components damaged.
Aside from that, I notice it is a crossfire Board, and can help but think maybe they lack some of the full NVidia support.
 
This too might make a difference these are footnotes to your MB:
 
(Note 2) For optimum performance, if only one PCI Express graphics card is to be installed, be sure to install it in the PCIEX16 slot.
(Note 3) When the PCIEX1_2 slot is populated with an expansion card, the PCIEX4_X1 slot will operate at up to x1 mode. 
(Note 4) The PCIEX16 slot operates at up to x4 mode when ATI CrossFireX is enabled.
 
Honestly, it doesnt sound like a spectacular motherboard given that using crossfire throws it down to x4 bus speeds.  And it makes me think more about its inability to power a gtx480 if it is in anyway damaged.

No the worst advice would be not to RMA Either. 
If you have friend perhaps nearby that has a system to test it in, I would do that first. You can easily rule out the Card that way versus some other issue. 

Thanks for the advice (and for taking my harsh criticism so well... I apologize for that).
 
I don't think it's the motherboard because my PC works fine with the GTS 8800. I can't use the 4x PCIe slot anyway since it would cover my SATA ports. I have another PC I could try, but that would require me installing a new PSU (I have one, but that's a lot of work :( ) 
 
I considered testing the card in the 4x slot, but again, the SATA ports get in the way.
 
Someone from nVidia contacted me on the other boards, so I think nVidia is actually looking into this. I'm going to hold off on doing anything for a couple weeks, anyway, since I'm in Europe. Hopefully nVidia has some good news by then.

Another Thought is just some bios setting that you are missing?
Maybe some form of power management? I know alot of the new settings on Motherboards are jumping on the Green Bandwagon to save energy and save the world. It could be some of these are activating and dropping power to the PCI. Just a wild theory. But personally I turn all of those power features off in BIOS and in Windows. I want my PC Running and humming 24/7 i dont want it getting lazy with my data, taking a nap, or dropping a connection.

It could be you just ran into bad luck, bios does not get along with card or you were one of the few who got a bad card. It happens after all in anything you buy I would call EVGA tbh and setup an rma.