Saturday, July 2nd 2016

Official Statement from AMD on the PCI-Express Overcurrent Issue

AMD sent us this statement in response to growing concern among our readers that the Radeon RX 480 graphics card violates PCI-Express power specification, by overdrawing power from its single 6-pin PCIe power connector and the PCI-Express slot. Combined, the total power budged of the card should be 150W, however, it was found to draw well over that power limit.

AMD has had out-of-spec power designs in the past with the Radeon R9 295X2, for example, but that card is targeted at buyers with reasonably good PSUs. The RX 480's target audience could face troubles powering the card. Below is AMD's statement on the matter. The company stated that it's working on a driver update that could cap the power at 150W. It will be interesting to see how that power-limit affects performance.
"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 8 Gbps 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)."
Add your own comment

358 Comments on Official Statement from AMD on the PCI-Express Overcurrent Issue

#226
ensabrenoir
...took an Am2 and a socket 775 to produce a crash in this video...



for the most part a fix is coming and as long as the fix don't kill performance... and they take care of those who's board got damaged........nothing else going on. Next story please!
.....pooka dots? Not promoting this guys taste in T-shirts.
Posted on Reply
#227
newtekie1
Semi-Retired Folder
Ungari
This idea of average load vs. power peak spikes argument is a diversion that favors Nvidia's lower average due to it's more extreme oscillations.
If you would agree that overclocking cards like the 750's and 950 SE exceeds the power specs for PCIE, then where are all the broken mainboards?
No, the average vs. spike argument is because an average load over time heats things up and can kill hardware. A spike doesn't heat up the connector and cause damage.

The 960 might spike to 225w, but a lot of cards spike pretty damn high, it is a byproduct of DC to DC conversion. The difference is the GTX960 overages significantly under spec on average, like only pulling ~30w from the PCI-E slot over time. So the connector doesn't heat up, and there isn't a risk of damage. The RX 480 on the other hand average way over the spec, which will heat things up and can cause damage.

The funny thing is, I've seen a 24-pin connector melt on a PSU with two 480s connected and folding...they just weren't AMD's 480, they were nVidia's. It wasn't exactly an ancient board either, it was a ASUS 990X based motherboard. And the PSU wasn't a slouch either, it was a Corsair HX850.
Posted on Reply
#228
cadaveca
My name is Dave
newtekie1
The funny thing is, I've seen a 24-pin connector melt on a PSU with two 480s connected and folding...they just weren't AMD's 480, they were nVidia's. It wasn't exactly an ancient board either, it was a ASUS 990X based motherboard. And the PSU wasn't a slouch either, it was a Corsair HX850.
Well, this is where the concerns should be placed... on those running multiple cards in a system. 16W on a single card becomes 48W on three cards. And since AMD doesn't have a 2-card limit to the number of cards in Crossfire configs, this sort of thing is something people who are considering such rigs need to consider.

This isn't a huge issue at all, however, it is something that needs to be considered depending on how many of these cards you might plan to use.
Posted on Reply
#229
newtekie1
Semi-Retired Folder
cadaveca
Well, this is where the concerns should be placed... on those running multiple cards in a system. 16W on a single card becomes 48W on three cards. And since AMD doesn't have a 2-card limit to the number of cards in Crossfire configs, this sort of thing is something people who are considering such rigs need to consider.

This isn't a huge issue at all, however, it is something that needs to be considered depending on how many of these cards you might plan to use.
I agree, this isn't a huge issue. But definitely an issue that needs to be addressed. The people that are downplaying this and saying the issue isn't really an issue are just plain wrong. It is causing problems. They are saying don't worry because newer motherboard have protection to stop damage. Ok, fine. But my machine just shutting down randomly is an annoying f'n issue that I'd want fixed. It definitely isn't "no real issue".

Plus, as with my experience with the GTX480s, the problem worsened over time until the connector completely melted. I didn't even realize what happened until it was too late. The machine ran fine for almost a year. Then started randomly shutting down about once a month. Then it started happening about once a week. I did visual inspections, all looked fine. I even wiped and re-installed Windows once it started happening once a week. Then one week it shut down every day for about 4 days straight. Then then finally it shut down and wouldn't power back on. It wasn't until I pulled the computer completely apart that I found the burn/melted 24-pin connectors on the PSU/Motherboard.

And that is the thing that concerns me, even with only running one card. Over time, if you are constantly over-driving the connector, it can deteriorate. And the problem with a power connector is, when it start to deteriorate it takes more current to overcome the resistance of the poor connector. More current and more resistance means more heat at the connector. It is just a snowballing affect until failure.

I'm not saying this is something that is going to happen all the time. But even something like dirty contacts cause cause more resistance and a higher potential for failure. So it is definitely a possibility. And there really isn't any reason to pull that much power from the PCI-E slot. The 6-pin/8-pin connectors are way overbuilt. The extra power should be pulled from that connector.

And obviously custom designed cards aren't likely to have this problem anyway.
Posted on Reply
#230
cdawall
where the hell are my stars
cadaveca
Well, this is where the concerns should be placed... on those running multiple cards in a system. 16W on a single card becomes 48W on three cards. And since AMD doesn't have a 2-card limit to the number of cards in Crossfire configs, this sort of thing is something people who are considering such rigs need to consider.

This isn't a huge issue at all, however, it is something that needs to be considered depending on how many of these cards you might plan to use.
To be fair most of the boards that would have 3 RX480's would have an add on connector on the board for more power to the PCI-e
Posted on Reply
#231
Prima.Vera
ensabrenoir
...took an Am2 and a socket 775 to produce a crash in this video...



for the most part a fix is coming and as long as the fix don't kill performance... and they take care of those who's board got damaged........nothing else going on. Next story please!
.....pooka dots? Not promoting this guys taste in T-shirts.
Really dude. I stopped on the second nr. 2 immediately after I saw the guy dressed like a woman...
Posted on Reply
#232
Ungari
It's funny that so many think that AMD should have put an 8-pin instead of a 6-pin, as if this would changed the power draw from the motherboard slot.
Posted on Reply
#233
Ungari
ensabrenoir
and they take care of those who's board got damaged
I haven't heard of any credible article where someone lost a mainboard due to the RX 480, have you?
Posted on Reply
#234
cdawall
where the hell are my stars
Ungari
I haven't heard of any credible article where someone lost a mainboard due to the RX 480, have you?
Only one I have seen was a user report of a dead asrock board, but it was well abused.
Posted on Reply
#235
AsRock
TPU addict
cdawall
Only one I have seen was a user report of a dead asrock board, but it was well abused.
As much as the 480 is in that last video some one posted, like OMG he only just got that lol.

Prima.Vera
Really dude. I stopped on the second nr. 2 immediately after I saw the guy dressed like a woman...
His problem not ours, still don't mean the info he found was not bad or good.

ensabrenoir
...took an Am2 and a socket 775 to produce a crash in this video...



for the most part a fix is coming and as long as the fix don't kill performance... and they take care of those who's board got damaged........nothing else going on. Next story please!
.....pooka dots? Not promoting this guys taste in T-shirts.
As i heard it he tested a 775 setup were as he said some one else tested the AM2.

But again OMG few days later and the card looked like this Sheesh.

Posted on Reply
#236
ensabrenoir
AsRock
As much as the 480 is in that last video some one posted, like OMG he only just got that lol.



His problem not ours, still don't mean the info he found was not bad.



As i heard it he tested a 775 setup were as he said some one else tested the AM2.

But again OMG few days later and the card looked like this Sheesh.


The card still had the plastic protection on it.....hes a responsible youtuber.
Posted on Reply
#237
AsRock
TPU addict
ensabrenoir
The card still had the plastic protection on it.....hes a responsible youtuber.
Youtubers the place you keep going lol.


EDIT, yes my bad :P.
Posted on Reply
#238
Ungari
ensabrenoir
The card still had the plastic protection on it.....hes a responsible youtuber.
He's definitely not riding dirty!
Posted on Reply
#239
ensabrenoir
.....wonder if we can strerch this out till tuesday or when ever they release their update. Then we can start a performance impact debate....well its late one trouble maker signing off.
Posted on Reply
#240
rtwjunkie
PC Gaming Enthusiast
NDown
Well what do you expect when most of their fanbase are mostly manchild/literal kid

you cant have a good gaming experience if you dont have the GeForce GTX® logo/sticker in your PC afterall :^)

most probably doesnt care about efficiency either, or they are simply too new to remember the HD5xxx vs GTX 4xx series
And exactly what are you doing? Both sides fanbois are nasty and cruel, and arguing about shit that just doesnt matter while they throw insults like you just did.

And for the record, most of TPU's membership is well over 30, with alot of us in our 40's and 50's.
Posted on Reply
#241
R-T-B
rtwjunkie
And for the record, most of TPU's membership is well over 30, with alot of us in our 40's and 50's.
I wouldn't say most. There are a lot more one post wonders that are probably little kids. But in post count, logic, and participation, us 30ish people hold our weight quite well.
Posted on Reply
#242
RejZoR
Dippyskoodlez
Did you not see my screenshot? My GTX 970 runs "346.03".

There are also Linux users using non-proprietary blobs that provide basic acceleration. If a system fires up any compute, you could run into problems, especially considering it's likely to beat any game power draw.

I don't have any evga 'gaming' software' to run, I don't have GPU-z to tweak things, I have strictly the on card BIOS to control cooling of the ACX2.0 fans. If it were driver controlled thinking it was a blower, the 'off' feature at low load/temperatures wouldn't function.

An RX480 trying to pull >75w through my pcie slot would actually start causing stability problems if it doesn't actually cause any physical mishaps from the overcurrent, as is very well documented in eGPU setups already.

Edit: On a side note, this is why the 'bullshit' reviewer bios fiasco is also very important to me. I can't necessarily take advantage of an overclock if it's not preset.
If you run Linux and "compute" you're not an average user and you've heard about RX480 "issue". And you don't seem to understand BIOS/driver relationship at all either. Drivers don't "assume" you have a blower cooler on your ACX EVGA graphic card. It KNOWS you don't have it. That's why AIB's have different hardware ID's to identify specific hardware, so that driver doesn't "assume" things like this, but it "knows" things like this.
Posted on Reply
#243
basco
pcper just gave us a second view of things on rx480 and i like the explanation and data.
maybe tomshardware+others should have not get it out so quickly and talked to other sites before making it big.
i think this is blown out of proportion.

rbuass is a well known and respected overclocker:
with subtitles

Posted on Reply
#245
Frick
Fishfaced Nincompoop
basco
maybe tomshardware+others should have not get it out so quickly and talked to other sites before making it big.
i think this is blown out of proportion.
/MEDIA]
The revenue must flow. - Guild Navigator
Posted on Reply
#246
john_
I wonder if GTX 960 Strix was brought as a paradigm from those defending Nvidia. It's a card with an extra power connector and TDP much lower than 150W. Whatever spikes it produces the average consumption from the pcie bus will always look under the limit.
But tell someone about an overclocked GTX 950 without a 6 pin connector and ignores you, changes the subject, or wants you to believe that you can have 20% extra performance without consuming a single extra watt. Free performance.


Instead of people start asking if there are other graphics cards out there depending too much on the pcie bus, or sites starting tests to see if under overclocking specific graphics cards could push the pcie bus over it's specs(the GTX 950 I mentioned, the R9 270X with only one pcie connector could be another example, GTX Titan Z, R9 295X2), this ends up again becoming the favorite subject for many people. How to attack a new product from AMD.

One more opportunity to attack AMD, a lost chance to investigate on something important and probably learn something we seemed to ignored until today and probably we will ignore it in the future, starting after tomorrow.

jigar2speed
Thought to share it with you guys...


Yeap. A fact that many try to hide behind their little finger.
Posted on Reply
#247
ArdWar
I can't even understand why peoples trying to, or experimenting if their motherboard would crash when used with reference RX480.

A power overload wouldn't crash a computer unless it's so grossly overloaded that it'll trip PSU OPP/OCP, or drop the PSU rail significantly under ATX spec. The PCI-E power delivery is just straight connected into power and ground plane on PCB, almost no other component inbetween to be directly affected other than the connectrors itself. There's no power limiter, power sharing controller or whatever on the motherboard itself. Just direct connection!

If anything, the problem wouldn't show itself so quick. A newly inserted power connector and PCI card is literally the best case scenario for power delivery. The contacts are still shiny new, no oxidation, no dirt, and the scratch created during card insertion would help with removing the already existent oxide layer. Wait some month/years till the contacts oxidated, increasing the contact resistance. By Ohm's law, heat generated by the current flowing in the contacts will increase and depends on the material, maybe it'll melt the plastics on the contacts.


A bad motherboard that skimp on using proper power plane and ground plane, using skinny traces instead, is another story. It could literally burn.
Posted on Reply
#248
ixi
AsRock
As much as the 480 is in that last video some one posted, like OMG he only just got that lol.



His problem not ours, still don't mean the info he found was not bad or good.



As i heard it he tested a 775 setup were as he said some one else tested the AM2.

But again OMG few days later and the card looked like this Sheesh.


That's racist!
Posted on Reply
#249
AsRock
TPU addict
ixi
That's racist!
No, that's just you taking shit out of context.
Posted on Reply
#250
ixi
AsRock
No, that's just you taking shit out of context.
Nice humour you got there mate.
Posted on Reply
Add your own comment