AMD RX 480 Fails PCI-E Specification

humbug · 1 Jul 2016 at 17:23

3 here with fatally damaged Motherboards, all of them older AMD boards, two 990FX boards and one 970 board. see comments.

https://community.amd.com/thread/202410

I thinking now this is a real problem, AMD need to get a handle on it fast.

TwsT · 1 Jul 2016 at 17:25

Yrkoon said:
the guy who is claiming it took out his pcie slot seems to have been overclocking the card.

"RX 480 8GB Overclocked to 1350MHz core (no overvolting, stock VRAM speed)"

that is in his sig.

hope it is not true though.

it didn't break either of my PCI-E slots when i overclocked to 1330 in xfire

http://www.3dmark.com/fs/8902886

Orcvader · 1 Jul 2016 at 17:26

humbug said:
3 here with fatally damaged Motherboards, all of them older AMD boards, two 990FX boards and one 970 board.

https://community.amd.com/thread/202410

I thinking now this is a real problem, AMD need to get a handle on it fast.

Might want to check a few posts up

.

Orcvader said:
Not sure if trolls or not, but some people on the AMD forums reporting dead PCIe slots: https://community.amd.com/thread/202410

humbug · 1 Jul 2016 at 17:28

Edit ^^^ nevernind..

Orcvader said:
Might want to check a few posts up .

Right, apologies to Orangy. tho i am linking directly to a user with issues and comments with issues.

drunkenmaster · 1 Jul 2016 at 17:30

D.P. said:
DM was adamant that Pascal would arrive 6 months after Polaris.

ICDP said:
Not true, like many here he was going on the leaked information and rumours we had available at the time. He wasn't adamant, he was making a best guess.

Similarly I could claim you were adamant that 1070 would cost £320.

http://forums.overclockers.co.uk/showpost.php?p=29574092&postcount=10866

We all guess wrong from time to time but most of us are doing so from rumours or info we have at the time. Bringing up the fact someone made a bad guess or assuming alleged info is accurate is not something we should use to get one over on that person. We could end up making ourselves look very petty when it is demonstrated we are just as guilty of being adamant but wrong.

DP likes to make things up, I had to point out over and over to Greg that I at no stage was claiming Pascal COULDN'T or WOULDN'T be out in April(it wasn't which isn't surprising) but that the specific rumours and the informations those specific rumours were based off were inaccurate. The information was proven false, thus the rumours based off them were false, that didn't mean Pascal couldn't be out in April but that rumour would still be false regardless.

I also never said Pascal couldn't be out within 6 months of Polaris, I said the rumours put AMD and the process with a 3-6 month lead on TSMC, as such this implied Nvidia would be 3-6 months behind AMD.

I also managed to point out that risk production could see EITHER company jump forward 3 months, Nvidia have done so previously and that AMD would almost certainly not have the cash available to take such a risk. On top of that I said risk production, paper launches and everything else aside the best way to compare is ignore the launches and see when both companies achieve similar shipping quantities because that is the best comparison of who is able to supply chips in large quantity and who is not.

The industry rumours still put 1080 as risk production, entering mass production LONG after AMD did with Polaris 11(supposedly around Jan/Feb). IF AMD are shipping 100k's of cards to OEMs and Nvidia are shipping a couple of batches of risk production 10-20k cards and didn't actually entire full scale production until mid May as appears likely, then Nvidia are still 3 months behind AMD. There are still continuing rumours of GP106 paper launching, not supposedly being widely available, it has no official launch date and Gp104 supply is horrendous.

AMD sold more or a single Polaris 10 SKU to retailers alone than Nvidia has managed to sell of two SKU's in a month. All of this points directly to AMD being in mass production significantly before Nvidia was.

Orcvader · 1 Jul 2016 at 17:31

humbug said:
Edit ^^^ nevernind..

Right, apologies to Orangy. tho i am linking directly to a user with issues and comments with issues.

But I posted the same link you did before Orangy's post :confused:

.

The one Orangy quoted looks like an obvious troll. The rest does seem a bit more realistic. Though they all have used older AMD boards, which may have not been able to supply the extra power demand.

humbug · 1 Jul 2016 at 17:40

Orcvader said:
But I posted the same link you did before Orangy's post .

The one Orangy quoted looks like an obvious troll. The rest does seem a bit more realistic. Though they all have used older AMD boards, which may have not been able to supply the extra power demand.

Looks like the author of that thread is revelling in the drama of it all, which is unfortunate but it is starting to look like the problem is real, if only for older boards.

flopper · 1 Jul 2016 at 17:53

drunkenmaster said:
AMD sold more or a single Polaris 10 SKU to retailers alone than Nvidia has managed to sell of two SKU's in a month. All of this points directly to AMD being in mass production significantly before Nvidia was.

Obviously hard to sell overpriced solutions.
16nm is pricey and if yields are bad your skyrocket the cost.
its likely 2-3 or so years before there is some sane price with yields.
the more cautios approach with the 480 and 470 designed for the low/mid market was a good idea.

the 480 passed the pci-e validation thats a fact.

drunkenmaster · 1 Jul 2016 at 18:07

I should also mention, what I was adamant about was that the rumour that lets not forget, stated Nvidia got test chips back for a Gp100 in November(iirc) because a list of parts shipped from Nvidia was found. This list of parts clearly showed the shipping of test equipment for chips that would naturally be received later, nothing more or less. Every single rumour quoted that list as test chips coming in for November and thus GP100 would be launched in April... a chip still afaik not shipping anywhere, a chip that won't even be sold as a gaming GPU anyway.

These rumours were proven BS at the time because the information they were based on was nuts, they were proven correct in March/April when Nvidia paper launched a GP100 they said would be available in Q1 2017 and GP104 launched in May with epically awful availability and now in July still has awful availability and tiny sales volume. However in releasing when they did... not just to be first or anything, they managed to tank their own pricing on their entire range which will have only reduced their profits. Tanking profits on 99% of your sales to launch a card early just to be first before AMD, well, it certainly implies Nvidia are late and they wanted their own card to be the reason for their short term reduced profits rather than AMD's Polaris line up being why Nvidia had to tank pricing across the majority of all their products being sold.

D.P. · 1 Jul 2016 at 18:16

ICDP said:
Not true, like many here he was going on the leaked information and rumours we had available at the time. He wasn't adamant, he was making a best guess.

Similarly I could claim you were adamant that 1070 would cost £320.

http://forums.overclockers.co.uk/showpost.php?p=29574092&postcount=10866

We all guess wrong from time to time but most of us are doing so from rumours or info we have at the time. Bringing up the fact someone made a bad guess or assuming alleged info is accurate is not something we should use to get one over on that person. We could end up making ourselves look very petty when it is demonstrated we are just as guilty of being adamant but wrong.

The 1070 would be much closer £320 if there wasn't a BREXIT. 379 RRP was around £320 with VAT, using the same conversion as the 1080 cards.

And no, DM was adamant that Pascal would be 6 months behind Polaris even when all the information indicated otherwise. All the rumours suggested Big Pascal announcement in April with 1080 using GDDR5X available in Q3. that is exactly what happened, as rumoured since Q3 2015.

The Running Man · 1 Jul 2016 at 18:33

meldarthx said:
nope you implied that several reviewers had their mbs blown up by 480s......so either show it......or don't post these things....

Presumably reviewers have top end cpu, mobs and PSu configurations unlike people who's buying a mid end card?

humbug · 1 Jul 2016 at 18:49

D.P. said:
The 1070 would be much closer £320 if there wasn't a BREXIT. 379 RRP was around £320 with VAT, using the same conversion as the 1080 cards.

And no, DM was adamant that Pascal would be 6 months behind Polaris even when all the information indicated otherwise. All the rumours suggested Big Pascal announcement in April with 1080 using GDDR5X available in Q3. that is exactly what happened, as rumoured since Q3 2015.

There were £380+ before Brexit, they are now £390+

iMik · 1 Jul 2016 at 18:51

Power Consumption Concerns on the Radeon RX 480

http://www.pcper.com/reviews/Graphics-Cards/Power-Consumption-Concerns-Radeon-RX-480

brigantium · 1 Jul 2016 at 19:05

Card should be removed from the shelves, it will spread like wildfire and hurt the retailers.

StrongSad · 1 Jul 2016 at 19:14

brigantium said:
Card should be removed from the shelves, it will spread like wildfire and hurt the retailers.

Seeing that the most traction I have seen on this issue (It was there from launch) is from this thread... Then no they should not.

It most likely only affects older MB and possibly older PSU's near the top of their output.

There is an RMA/returns policy for new kit from most suppliers and if there is an issue then they can be returned or offered an AIB partners card when available with additional power. What I do find mad is that the reference only has one 6 pin....

harney · 1 Jul 2016 at 19:17

The Running Man said:
Presumably reviewers have top end cpu, mobs and PSu configurations unlike people who's buying a mid end card?

yep dread to think some of the systems this rx 480 will be going in
low end psu's motherboards ect...but surly there would be far more reports if it was a real issue or is it that its just begun ....hope not as i want AMD to do well will see

harney · 1 Jul 2016 at 19:20

Strudders said:
Seeing that the most traction I have seen on this issue (It was there from launch) is from this thread... Then no they should not.

It most likely only affects older MB and possibly older PSU's near the top of their output.

There is an RMA/returns policy for new kit from most suppliers and if there is an issue then they can be returned or offered an AIB partners card when available with additional power. What I do find mad is that the reference only has one 6 pin....

I think some one mentioned before that the RX 480 is now pushing more than it was originally designed for...

SkeeterUK · 1 Jul 2016 at 19:30

Strudders said:
Seeing that the most traction I have seen on this issue (It was there from launch) is from this thread... Then no they should not.

It most likely only affects older MB and possibly older PSU's near the top of their output.

There is an RMA/returns policy for new kit from most suppliers and if there is an issue then they can be returned or offered an AIB partners card when available with additional power. What I do find mad is that the reference only has one 6 pin....

Isnt the whole point of this 480 card was to be a upgrade for people who cant afford 300+ cards so and im just guessing here but a lot of customers who buy it could be using old hardware from 5 years or more ago and probably will be using cheap motherboards.

hyperseven · 1 Jul 2016 at 19:38

iMik said:
Power Consumption Concerns on the Radeon RX 480

http://www.pcper.com/reviews/Graphics-Cards/Power-Consumption-Concerns-Radeon-RX-480

would someone kindly explain those amp drop spikes?

not my area of expertise :confused:

I am genuinely interested.

D.P. · 1 Jul 2016 at 19:41

hyperseven said:
would someone kindly explain those amp drop spikes?

not my area of expertise

I am genuinely interested.

The voltage is dropping because the motherboard and PSU are struggling to provide the power demands of the 480 since it is going well above spec.