From what I've gathered from the post if you OC it you run the risk of drawing to much power from your motherboard. Which after awhile could fry your PCI slot if you are lucky. I'd watch out for this and at the very least not OC the card at the moment.
edit 5: Seems like AMD has posted a driver update that supposed to shift some of the power load off of the PCIe slot in order to bring it within specification, however, this does put your 6 pin out spec... So people with old/crappy PSU probably should avoid OCing the GPU, but AMD has setting for you in the Radeo settings menu in order to fix this. Anandtech also ran some performance checks on it so I'd definitely check out the link.
Your gpu might not have enough power and loose stability.
other issues is when hot air blows on ports (some of coolers like msi's gaming one) over prolonged time can melt the plastic port. (happened a lot with lga 2011 mobo's since ram slots were really close to pci-e bus - but it wasn't fault of gpu... just poor design of mobo)
those type kind of mobos had this problem
as you can see pci-e port is pretty much beside 4 memory slots... well easy to melt pci-e port or memory slots... but rx480 is too easy ... you'd need something that uses at least 400W
Ok, did some research. Came up with a maximum of 300W, but that's the max. The source I found wasn't from a PC manufacturer or a education site. From what I gathered it's more or so up to the PC manufacturer to support this. So it's dependent on the motherboard manufacturer/chipset....
edit: However this is for PCI 3.0 so if you're using 2.0 you probably run the risk of card running under powered or damaging the mobo.
but is this 75 watt total sustained power from turning on to turning off, or is ist for 10min of load, or ... no one ever asked for the observation period.
It is kind of overblown but it still might be a problem if you have an older motherboard. The solution is to wait for aftermarket models with an 8pin connector. You should never buy reference models anyway.
I'll tell you when my mobo melts. Currently -15% power target set and <100 W per card being drawn. Less than 600W total system power. All cards running below 70 degC fans ~3000RPM. That may go up as I learn how to optimize the current setup.
75W sustained is PCIe spec. That is 3.3V and 12V combined. Spikes up to 90W or even more are tolerated on nearly all boards. But more than 75W sustained might lead to problems over the long term.
You realize the card draw may be 80 watts instead of 75 watts, right? You can lift up 120kg, and if you lift 130, you don't die of exhaustion immediately. The thing is, you may draw 80-85W from the PCI slot once in a while. I doubt it eats constantly 85W... You play 16 hours nonstop? I don't think so.. 1070 also eats more power than normal, nobody talks about that shit... It is an issue, it could be solved easily... If you want a GPU NEVER buy reference... Get an aftermarket card, like the Sapphire Nitro i am hyped about, or whatever Powercolor bring to the market... God forbid ASUS and MSI actually market a card, that sells like 20 times more than what they heavily market... So i don't care...
Don't worry I just ran my tongue along the ribbon cable and it didn't get burned!
Maybe gaming would be a different story but you have to be nuts to run 4 way GPUs in gaming now .... though maybe I'll try over the weekend just for kicks!
The 1070 has an 8 pin. So it can draw more power from there. 6 pin is limited to 75W, slot is limited to 75W. Basic math dictates that a power draw of 166 is not good. Don't get me wrong, I still think most new boards will be fine but there are old boards, budget boards which won't be fine. Here is an example:
AMD should fix this ASAP or they might ruin the reputation of this card.