RX 480 PCIE Problem

[Megazero]

So this seem like a much bigger problem than I initially thought, so I will be posting this on the News section

 

To summary the problem, the RX 480 as all other GPU pull power from both the 6 pins connector and the PCIE on the motherboard. The problem is, the power it pull from the PCIE (which by specification shouldnt pull more than 66W) exceeded the limitation/specification of the PCIE standard (it actually pull up to 200W peak, 100W average), making it can potentially damage the motherboard (or outright blow it), especially the low end one, which lots people who concern about budget might have given that they purchase this GPU.

 

(Source for the actual 100W 200W number: http://www.tomshardware.com/reviews/amd-radeon-rx-480-polaris-10,4616-9.html, and Quote:

Quote

Radeon RX 480’s power consumption through the PCIe slot jumped to an average of 100W, peaking at 200W

)

According to Jay (who kinda quote a lawyer on Reddit it seem), because the card have the "CF certified" and "PCIE" logo, it mean that it is certified to follow the standard of those things. But since it clearly dont as we have this problem, they can no longer put that logo on the box, "making it NOT a PCIE card". And there's 2 option for AMD
 

Quote

* Do a mass recall of ALL the graphic cards that are sold, add a 8 pin and fix the problem
* Keep the logo and deal with a lawsuit

Also according to Jay this can be a problem that even the AIB card cant fix. CAN, not absolute yet. The info is still kinda low on this right now but that's also something we need to keep in mind

So... what's everyone thought about this?

[shadowbyte]

@byalexandr

please explain 750ti issue

[NumLock21]

 

[maulemall]

I think ... No I can, With 100% certainty say,  Yes.

[byalexandr]

2 minutes ago, shadowbyte said:

@byalexandr

please explain 750ti issue

Basically the 750 Ti had this issue but has it far worse. The RX 480 'potentially' exceeds PCI-E compliance while the 750 Ti in most cases did. By far. Like it says in that podcast, the 12V rail supplies 66W for the PCI-E power. The 750 Ti, without any sort of overclock (even without a boost clock), draws up to 70W at peak load, around 65W at gaming load, again with no OC.

 

Now to me that seems a bit ridiculous when you compare it to this RX 480 situation. The 480 'might' exceed it, AMD is still testing, but the 750 Ti clearly does, sold a lot more, and is a lot older. And no one even batted an eye, because they have nothing to hate on because it's not from AMD and everyone has to hate AMD for some reason.

[Pickelhaube]

I rate this 75/66

[Deli]

Class action lawsuit coming?

[Megazero]

2 minutes ago, shadowbyte said:

@byalexandr

please explain 750ti issue

I just read his top post after clicking on his name, here's 2 problem with comparing this to a 750Ti. @byalexandr

1. It draw 20W more. This draw like close to 40W on average more and peaked at ~ 140W more. Especially when the more u OC the more it draw from PCIE instead of the pin it seem. Thought I do agree that "potentially" vs "actual" is like you said, but 
2. As initially thought I also didnt think this is such a big problem. BUT after Jay talk about the actual legal implication of this whole issue it could actually fair much worse.

[Megazero]

2 minutes ago, byalexandr said:

Basically the 750 Ti had this issue but has it far worse. The RX 480 'potentially' exceeds PCI-E compliance while the 750 Ti in most cases did. By far. Like it says in that podcast, the 12V rail supplies 66W for the PCI-E power. The 750 Ti, without any sort of overclock (even without a boost clock), draws up to 70W at peak load, around 65W at gaming load, again with no OC.

 

Now to me that seems a bit ridiculous when you compare it to this RX 480 situation. The 480 'might' exceed it, AMD is still testing, but the 750 Ti clearly does, sold a lot more, and is a lot older. And no one even batted an eye, because they have nothing to hate on because it's not from AMD and everyone has to hate AMD for some reason.

It's not hate that coming from people like Jay and Tom, it's more like "concern". I surely cant say I remember the problem with the 750Ti being reported, but I sure remember how every bashed Nvidia for the whole 3.5GB fiasco.

[Megazero]

7 minutes ago, NumLock21 said:

 

Oh LOL. I checked 2 pages and dont see it so I posted a new one. If it already existed then a mod should closed this topic then

[byalexandr]

1 minute ago, Megazero said:

It's not hate that coming from people like Jay and Tom, it's more like "concern". I surely cant say I remember the problem with the 750Ti being reported, but I sure remember how every bashed Nvidia for the whole 3.5GB fiasco.

They did receive a shit storm for that.

 

If the 480 does exceed the compliance then at the least I would hope for a recall that changes the 6 pin to an 8 pin. Also keep in mind that although the 480 draws more power, it has the 6 pin by default whereas the 750 Ti did not.

[Megazero]

1 minute ago, byalexandr said:

They did receive a shit storm for that.

 

If the 480 does exceed the compliance then at the least I would hope for a recall that changes the 6 pin to an 8 pin. Also keep in mind that although the 480 draws more power, it has the 6 pin by default whereas the 750 Ti did not.

That's another thing that concern me. Like Jay said that if the core somehow send command to draw power from the PCIE instead of the pin, board partner adding in addition power pins may not be able to fix this. So the issue may not just be simply adding more pins to the card and may require some bigger change. If they fucked up as hard as to have to completely redesign the PCIE, then... Hopefully it will never reach that point

[awesomeness10120]

Someone on Pcper brought up a great point. The GTX 960 had the same problem, but no talked about that: http://www.tomshardware.com/reviews/nvidia-geforce-gtx-960,4038-8.html

 

[Deli]

Is it possible to have some sort of driver update to move the power draw from the PCIe to the six pin? The six pin can handle a lot more than 75w.

[Megazero]

1 minute ago, Deli said:

Is it possible to have some sort of driver update to move the power draw from the PCIe to the six pin? The six pin can handle a lot more than 75w.

I recall seeing someone said that AMD have confirm that this cant be fix by simply driver/bios update, but dont quote me on that. If they can it will surely be alot easier.

[Megazero]

BTW @Ryan_Vickers please close down this topic, I didnt notice the other one already been open

[Guest Kloaked]

15 minutes ago, byalexandr said:

Basically the 750 Ti had this issue but has it far worse. The RX 480 'potentially' exceeds PCI-E compliance while the 750 Ti in most cases did. By far. Like it says in that podcast, the 12V rail supplies 66W for the PCI-E power. The 750 Ti, without any sort of overclock (even without a boost clock), draws up to 70W at peak load, around 65W at gaming load, again with no OC.

 

Now to me that seems a bit ridiculous when you compare it to this RX 480 situation. The 480 'might' exceed it, AMD is still testing, but the 750 Ti clearly does, sold a lot more, and is a lot older. And no one even batted an eye, because they have nothing to hate on because it's not from AMD and everyone has to hate AMD for some reason.

75 watts is the maximum a graphics card should be drawing from a PCIe slot. 

[Megazero]

11 minutes ago, Deli said:

Is it possible to have some sort of driver update to move the power draw from the PCIe to the six pin? The six pin can handle a lot more than 75w.

Jay confirm that AMD rep has confirm that it cant be fix by bios update ( around 16:00 mark in the video)

[Deli]

1 hour ago, Megazero said:

Jay confirm that AMD rep has confirm that it cant be fix by bios update ( around 16:00 mark in the video)

Well then, AMD has a problem.

[Guest Kloaked]

4 minutes ago, Megazero said:

Jay confirm that AMD rep has confirm that it cant be fix by bios update ( around 16:00 mark in the video)

Shocker.

[DarkBlade2117]

3 hours ago, Megazero said:

Jay confirm that AMD rep has confirm that it cant be fix by bios update ( around 16:00 mark in the video)

As much as I hate this ego-tistic bastard, he has a good point in this vid.

Again, about the 750Ti issue. I don't think it is as big as a deal as everyone says. I had my 750Ti OCd quite high probably pushing it near the 100w mark and ran it on a 350w HP PC and a shitty HP mobo. Yeah, it is bad but almost for certain AIB will make it a 8 pin connector and I am 99% sure that 6+2 pins can supply 150w of power.

[Hazorazor]

If any problems pop up I hope mistakes are paid for at the very least... Looking forward to using mine but this doesn't make me feel very safe in using it. They should've used an 8 pin. Also wouldve been nicer as most cables are 6+2 pins.

[rattacko123]

1 hour ago, awesomeness10120 said:

Someone on Pcper brought up a great point. The GTX 960 had the same problem, but no talked about that: http://www.tomshardware.com/reviews/nvidia-geforce-gtx-960,4038-8.html

 

That is even worse than the 480 O_O, although to be fair it is more of an issue with the Asus AIB version. The problem with the 480 is that the reference version designed by AMD has this problem. AMD is in a bad financial position unlike ASUS, so this could be a serious problem (in financial terms) if a lawsuit does occur

[rattacko123]

 

1 hour ago, DarkBlade2117 said:

Again, about the 750Ti issue. I don't think it is as big as a deal as everyone says. I had my 750Ti OCd quite high probably pushing it near the 100w mark and ran it on a 350w HP PC and a shitty HP mobo. Yeah, it is bad but almost for certain AIB will make it a 8 pin connector and I am 99% sure that 6+2 pins can supply 150w of power.

it's probably fine if you run 1 or two of these graphics cards, as it doesn't exceed the maximum possible power output of the entire motherboard. If you fill up all the pcie slots with more power hungry devices (such as more GPUs) this could potentially be a problem. Besides, people who buy this graphics card wont be using that many pcie cards in the first place, so it is most likely safe.

[Megazero]

1 minute ago, rattacko123 said:

 

it's probably fine if you run 1 or two of these graphics cards, as it doesn't exceed the maximum possible power output of the entire motherboard. If you fill up all the pcie slots with more power hungry devices (such as more GPUs) this could potentially be a problem. Besides, people who buy this graphics card wont be using that many pcie cards in the first place, so it is most likely safe.

On the contrary, Since AMD advertise this card in CF against a 1080, I think quite a few will use 2 of this and damage the board even more