AMD RX 480 Fails PCI-E Specification

SkeeterUK · 4 Jul 2016 at 15:41

I dont think he bought it from ocuk as he mentioned in post 267.

harney · 4 Jul 2016 at 16:24

http://ht4u.net/reviews/2016/amd_radeon_rx_480_review/index12.php

Kaapstad · 4 Jul 2016 at 16:29

pandem0nium said:
Did you go for four again?

Yes.

D1craig · 4 Jul 2016 at 17:39

montymint said:
the power draw 'issue' doesn't need to be a reason for return under the long distance trading act.

As long as he has all of the packaging and the card is in good condition, OCUK should refund it.

Once funds have cleared, there is nothing stopping the guy ordering an AIB

i wasnt saying he shouldn't be able to replace the card or return. i was saying the sellers should not be entertaining this as a reason to exchange it.

im sure he is in the 14 day period so he doesnt need to say this is why anyway.

SouthEastBlue · 4 Jul 2016 at 20:16

I'm within 14 days of course, only got the card on Thursday last week. Was from a different retailer starting with 'E' ending with 'R'. would have got from OCUK but for some reason couldn't see the RX480's on OCUK's website using search function.

Will see what tomorrow brings with this driver update otherwise I will send it back - happy to foot the postage bill either way.

southernorth · 4 Jul 2016 at 20:39

SouthEastBlue said:
I'm within 14 days of course, only got the card on Thursday last week. Was from a different retailer starting with 'E' ending with 'R'. would have got from OCUK but for some reason couldn't see the RX480's on OCUK's website using search function.

Will see what tomorrow brings with this driver update otherwise I will send it back - happy to foot the postage bill either way.

I'm in the exact same situation with you, only I bought mine elsewhere. The various reports on the successes of undervolting cards do look promising though.

On a more general note, if hypothetically, my 480 did still ruin my motherboard does anyone know how easy it would be to get compensation, and who would I be needing to contact?

Rroff · 4 Jul 2016 at 20:42

You wouldn't get any kind of compensation unless you could prove beyond a shadow of doubt that the GPU killed it and even then it would be an uphill battle.

LoadsaMoney · 4 Jul 2016 at 20:45

AMD’s power problem is a problem after all

Company burning the midnight oil

AMD is burning the midnight oil fixing a problem on its new Polaris card which it thought was not an issue last week.

Last week reviews started complain that the Radeon RX 480’s power draw was enough to fry lower priced motherboards. AMD denied it saying that the card had been signed off by all the important standards crowd.

AMD fanboys claimed that it was all part of a conspiracy by Nvidia fanboys to damn the card to oblivion. We were inclined to agree, given that only one reviewer had noticed the problem, and it could easily have been a card issue.

But then the reviews started to confirm the problem – the 150-watt card used on average 168W under load. The card drew 90W over the motherboard’s PCI-E slot, far exceeding the 75W maximum the slot is rated for.

German site Golem.de observed identical behaviour, as did TecLab and PC Perspective. PC Perspective discovered that AMD’s card drew 7 amps over the PCI-E slot’s +12v rail, which is rated for 5.5 amps maximum. The Radeon RX 480 also overdraws the board’s solitary 6-pin power connector.

While 95W spikes through the PCI-E connector can be handled by most motherboards sustained power levels that high could potentially fry a motherboard’s pins and connectors.
AMD says a software-based fix is incoming. It said that it has identified select scenarios where the tuning of some RX 480 boards was not optimal.

“As you know, we continuously tune our GPUs in order to maximize their performance within their given power envelopes and the speed of the memory interface, which in this case is an unprecedented 8Gbps for GDDR5. Recently, we identified select scenarios where the tuning of some RX 480 boards was not optimal. Fortunately, we can adjust the GPU's tuning via software in order to resolve this issue. We are already testing a driver that implements a fix, and we will provide an update to the community on our progress on Tuesday (July 5, 2016).”

If the software fix involves reducing the power draw for the RX 480, that’ll most likely lower performance in the reference editions of the card as well, especially since Polaris’ performance is closely tied to the card’s power limit.

This all could’ve been avoided if AMD had simply used an 8-pin or 2x 6-pin power connectors. If it does fix all this smartish it could have a really good card on its hands.

http://www.fudzilla.com/news/graphics/41032-amd-s-power-problem-is-a-problem-after-all

southernorth · 4 Jul 2016 at 21:00

LoadsaMoney said:
http://www.fudzilla.com/news/graphics/41032-amd-s-power-problem-is-a-problem-after-all

So, basically, this is potentially AMD's most disastrous product launch ever and the issue is unlikely to ever be fully fixed for original 6-pin cards.

GG AMD then.

thebennyboy · 4 Jul 2016 at 21:09

I trust Gibbo more than some random reviewers on the internet (one of which looks like he is having a heart attack whenever he has to say something pro-AMD). Yet to see any killed motherboards, or burnt out boards as all the naysayers keep saying.

AMD problem = Major problem
Nvidia problem = Small problem quickly tucked under the carpet

All this BS from both camps is doing my head in.

Rroff · 4 Jul 2016 at 21:19

southernorth said:
So, basically, this is potentially AMD's most disastrous product launch ever and the issue is unlikely to ever be fully fixed for original 6-pin cards.

GG AMD then.

I dunno about that - the power delivery side of it in theory should be fixable - I find it curious that these cards use so much power and struggle to even do 1500MHz at a mere 232mm2 though on 14nm FF+ - that is nothing short of disastrous. I hope it is just that the process is lagging a bit behind TSMC.

JediFragger · 4 Jul 2016 at 21:29

Downclocking Ahoyyyyyyyyyyyyy!!!

southernorth · 4 Jul 2016 at 21:29

thebennyboy said:
I trust Gibbo more than some random reviewers on the internet (one of which looks like he is having a heart attack whenever he has to say something pro-AMD). Yet to see any killed motherboards, or burnt out boards as all the naysayers keep saying.

AMD problem = Major problem
Nvidia problem = Small problem quickly tucked under the carpet

All this BS from both camps is doing my head in.

Yeah I agree, it's not a clear cut problem and hasn't been looked at objectively. I am still a bit surprised with Nvidia's offering, and why on earth they decided on 3/6GB. If they went 4/8GB and reduced prices to compete, I could have seen them destroy one of AMD's most hyped releases ever, but they didn't.

D1craig · 4 Jul 2016 at 21:31

southernorth said:
I'm in the exact same situation with you, only I bought mine elsewhere. The various reports on the successes of undervolting cards do look promising though.

On a more general note, if hypothetically, my 480 did still ruin my motherboard does anyone know how easy it would be to get compensation, and who would I be needing to contact?

SouthEastBlue said:
I'm within 14 days of course, only got the card on Thursday last week. Was from a different retailer starting with 'E' ending with 'R'. would have got from OCUK but for some reason couldn't see the RX480's on OCUK's website using search function.

Will see what tomorrow brings with this driver update otherwise I will send it back - happy to foot the postage bill either way.

i seriously dont get it. you are both complaining of a problem that isnt a problem. almost all other cards that have ever been made use more than the "limit" of the pcie lane.

what are you hoping to see different than what you have now? how do you even check how much power is going through your pcie lane?

dits · 4 Jul 2016 at 21:32

southernorth said:
Yeah I agree, it's not a clear cut problem and hasn't been looked at objectively. I am still a bit surprised with Nvidia's offering, and why on earth they decided on 3/6GB. If they went 4/8GB and reduced prices to compete, I could have seen them destroy one of AMD's most hyped releases ever, but they didn't.

Doesn't matter ppl will be buying the 6Gb version like mad. OC will be raking it in

southernorth · 4 Jul 2016 at 21:36

D1craig said:
i seriously dont get it. you are both complaining of a problem that isnt a problem. almost all other cards that have ever been made use more than the "limit" of the pcie lane.

what are you hoping to see different than what you have now? how do you even check how much power is going through your pcie lane?

Look into it more and you'll see why it is more serious than you think.

Rroff · 4 Jul 2016 at 21:43

D1craig said:
i seriously dont get it. you are both complaining of a problem that isnt a problem. almost all other cards that have ever been made use more than the "limit" of the pcie lane.

what are you hoping to see different than what you have now? how do you even check how much power is going through your pcie lane?

None of those cards draws more than the limit from the PCI-e socket itself AFAIK while the 480 has been proved to be capable of drawing upto something like 92 watt sustained load which is around 40% over the spec.

Aside from that you have 3x 0.7mm traces carrying 12v from the socket which I believe are 1.37mils (might be wrong on that) - which gives a max current rating of ~5.46 amps - hard science means things don't go well when you start going much over that and IIRC the 480 has been demonstrated at 7.6amp draw from the socket - to pretend there isn't a potential problem there is naive at best.

EDIT: I forgot to include temperature range here and calculated it against the base temperature - the upper end of acceptable does actually allow for around 7-8 amps though that leaves no margin for environments outside of "normal" ones.

harney · 4 Jul 2016 at 22:05

Rroff said:
None of those cards draws more than the limit from the PCI-e socket itself AFAIK while the 480 has been proved to be capable of drawing upto something like 92 watt sustained load which is around 40% over the spec.

Aside from that you have 3x 0.7mm traces carrying 12v from the socket which I believe are 1.37mils (might be wrong on that) - which gives a max current rating of ~5.46 amps - hard science means things don't go well when you start going much over that and IIRC the 480 has been demonstrated at 7.6amp draw from the socket - to pretend there isn't a potential problem there is naive at best.

Yes this ^ ...
more peeps seem to be babbling on saying its all fan boy stuff ..its not well there riding it like that so its fuel for there fires but its all rather childish ... its a very serious problem this ...

the problem being is it may not cause problems straight away but over time it will eventually weaken the mb traces and the 12v power plugs to the board the connections are the weak points here and its all down to heat and time.

All so with the other cards 750 960 ect there time is very rapid spikes so the average is low pulling from the PCI slot..the 480 seems to sustain drawing longer more spikes so the average is much higher ....

pulling over 7.6 amps when the plug connectors are rated at 5 amps is a no no in my book ....same applies for the MB traces lanes one only hopes that the rated amps from the sockets and lanes is under spec so there could be room to play .....

the question is as damaged been done all ready weakened boards and sockets time will tell ....

fastboy · 4 Jul 2016 at 22:23

if you ever want to hear the truth about a product,good or bad..listen to someone who doesn't have a vested interest in that product.

whenever I hear in 1 ear that there are no problems by a salesman,and I hear the opposite in the other ear from an independent party..i know who I believe.

Griffildur · 4 Jul 2016 at 22:24

fastboy said:
if you ever want to hear the truth about a product,good or bad..listen to someone who doesn't have a vested interest in that product.

whenever I hear in 1 ear that there are no problems by a salesman,and I hear the opposite in the other ear from an independent party..i know who I believe.

that sounds like common sense to me!