I dont understand the news about the rx 480 drawing more than 75 watts. Don't most super expensive high end gpus pull more than that as well? What am i missing
>>55382164
Thats not the issue, the issue is that its pulling more than that through the slot.
Full length PCI-E slots are specced to deliver up to a maximum of 75W. Any more than that and the increased resistance losses from moving through the board traces and across the slot conductors to the card will eventually cause thermal damage to both the card and the board.
Additionally, if the board doesnt have some form of PCI-E booster power socket and/or that power socket is not occupied, it can overstress the 12v power lead on the main 24-pin connector feeding the PCI-E slots, also resulting in overheating and board damage.
What instead should have been done and been happening is that the RX480 should have come equipped with a PCI-E 8pin power plug, and have the card focus on pulling most of its power through it with the remainder coming from the slot.
>>55382164
The stock AMD card only has one Sox pin power connector. A six pin connector can only give 75 watts. So, the card has to take power through the PCIe slot from the motherboard. the card is pulling too much power from the motherboard, and so in some cases, the motherboard will shut down.
>>55382164
>Don't most super expensive high end gpus pull more than that as well?
The problem is it draws over 75W through the PCIE slot, which is only rated for 75W.
This can damage your motherboard (already reports of this happening to some people) or just make the card not work at all.
Even expensive GPUs don't do this, AMD just fucked up bad with the 480.
Should I go ahead and preorder a Nitro 480 from overclockers uk? I'm actually in burgerstan.
I assume AMD is fully aware of this and just does not care?
>>55382575
They're working on a fix now. Appearantly the issue can be fixed in drivers.
There is no issue, literally ignore all the shitheads, nvidias graphics cards have done this for YEARS, but it happens one time for AMD and suddenly OMG LOOK HOW BAD IT IS, WE KILLED A FOXCONN MOTHERBOARD WITH IT.
because when I think of reliability, I think fucking cuckscon.
nvidia GPUs spike to over 250w and OMG, CONSTANT 78W? THREE WHOLE WATTS OVER PCIE SPEC? what about the nvidia cards pulling 175W? oh, that doesn't matter, its not AMD.
>>55382554
Don't buy any 480 until they fix this
1060 it is lol
>>55382621
This is not true, there has never been a card that did this other than the 480
That's why it's such a big deal, they really fucked up bad with the 480
>>55382661
But the 8pin connector on the Nitro will fix this if Drivers don't do it first.
why does it use so much power but can't put up the numbers?
>>55382689
Sadly we don't actually know that. The board could easily still draw over the limit on the PCIE slot if it was based on the reference design.
>>55382164
>Don't most super expensive high end gpus pull more than that as well?
They come with slots that allow them to eat more than 150W.
The RX480 only has slots that allow it to eat 150W, but the card uses 150-170W (especially if you overclock it). So it eats more charge than what the connectors are rated for, and this can cause them to get damaged.
>>55382712
it uses tech from 2012 GCN1.0
no idea why its so slow and hungry if its on 14nm though
heat is the bigger issue though
>>55382621
Because watts isn't watts.
High ampere drain is severely dangerous to components, while high voltage drain for equal watt is the standard.
>>55382859
Heat is an issue because AMD chose to cheap out on the heatsinks. The VRMs have poor cooling and the main heatsink is a finned block of extruded aluminum with a copper insert where the die makes contact. I wouldnt have minded the extra $10 it would have taken to use a proper heatpiped fin heatsink or vapor chamber heatsink.
Personally, I'm waiting for the AIB's to release their own cards before I consider getting one to replace my very aged GTX 660.
>>55382712
Its a factory overclocked card.
It only has 32 ROPs so it doesn't have the pixel throughput of Hawaii/Grenada.
They made Polaris10 a cheap and relatively simple layout, then clocked it as high as possible at the expense of power consumption to hit their performance target as cheaply as possible.
>>55382859
You're a complete retard.
>>55383039
I'm trying to decide between the sexy Nitro version from Sapphire, or the Devil version from PowerColor myself.
>>55383086
I personally want the Sapphire card, cause goddamn thats a sexy beast.
It also comes with DVI-D, which is good since my cheapass main monitor uses that.
I don't understand it either.
If mainboard allows more than 75W on the PCI-E then it's the *mainboard manufacturer's fault*.
My MOBO refuses to provide anything over 75W. If anything demands more than that (heavy OC) it will go unstable and crash. As expected.
Want more power? Sorry! You're getting cut off, 75W is the limit. Deal with it.
So what the fuck is this argument about?
You should be complaining about MOBO manufacturers going for cheapest, non-standard compliant designs.
A video card literally *CANNOT* control the power throttling mechanisms of PCI-E slot. That's 100% mainboard circuitry.
How retarded do you have to be to say it's a graphics card fault?
This has to be another one of those NVidia shit-flinging campaigns.
>>55383268
It is 100% the graphics card fault.
There are no other graphics cards that pull over spec on pcie, the 480 is the only one that does it.
Motherboard manufacturers are not going to design around one card fucking their shit up, the card is supposed to draw power within limits.
>>55383317
>Motherboard manufacturers are not going to design around one card
They have to ensure the wattage passing through their PCI-E slot is constrained by 75W.
Otherwise *their* parts are not standard compliant.
Are you seriously saying "motherboard manufacturers shouldn't comply with standards, because all it does is makes card not overdraw"?
Because that's how it looks.
And no. They *have to* comply with standards, no matter what.
>>55383268
No motherboard on the market has any kind of circuitry in place that hard limits power delivery in this regard. You're completely full of shit.
The issue is stemming from the memory modules on board drawing power with an absurd margin stemming from an issue with the Wattman program AMD just launched. Its getting fixed after the holiday weekend.
Do they all do this or just the reference card? I ordered the ASUS one since it was in stock and I needed a new card, but now I am minorly paranoid especially since I have a pretty shitty mobo I cheaped out on.
>>55383422
That's like saying roads should be made to fit cars that are 200ft wide, because there is one car that size.
The motherboards ARE compliant. The 480 is not.
>>55383479
All of them you can buy right now do it.
>>55383317
Strix 960 pulls over spec on PCI-E and does it much more consistently.
>>55383450
>No motherboard on the market has any kind of circuitry in place that hard limits power delivery in this regard.
And yet - the standard asks for it.
Funny, isn't it.
>>55383512
No it does not, see: http://www.pcper.com/reviews/Graphics-Cards/Power-Consumption-Concerns-Radeon-RX-480/Evaluating-ASUS-GTX-960-Strix
>>55383512
you're a lying shill faggot
>>55383520
No the standard asks for the pcie card to limit their power draw, not the motherboard.
>>55383550
Just like the power outlet standard asks appliances to not overdraw.
Gee, I wonder why do we bother with fuses then!
>>55383873
Again, the strix is well within limits, read this: http://www.pcper.com/reviews/Graphics-Cards/Power-Consumption-Concerns-Radeon-RX-480/Evaluating-ASUS-GTX-960-Strix
