The brand new AMD RX 480 seems to have an issue of drawing too much power through the pci-e port, so you may want to hold off on buying one of the first lot for now.
To make it clear, I cannot verify it myself since I have no contacts at AMD to get hold of a card.
Wccftech has the full lowdown here. As do a number of other respected websites.
It's worth noting that the Asus GTX 960 Strix also exhibits similar problems.
The main thing to note is: Don't overclock the reference AMD RX 480 as it could cause issues, but the custom designs from partners like Sapphire, XFX, Asus & MSI shouldn't have the issues.
The reference design also seems to be at the power limit, so hopefully cards from their partners will have additional power connections to feed it.
Still, this is a bit sad to see for a brand new card launch and something that's important for AMD.
This is why I think it's always a bad idea to be an early adopter, it reminds me of how Nvidia stung me with the VRAM issue on the 970 which now sits in my Steam Machine (I now have a 980ti in my desktop). Article taken from GamingOnLinux.com.
To make it clear, I cannot verify it myself since I have no contacts at AMD to get hold of a card.
Wccftech has the full lowdown here. As do a number of other respected websites.
It's worth noting that the Asus GTX 960 Strix also exhibits similar problems.
The main thing to note is: Don't overclock the reference AMD RX 480 as it could cause issues, but the custom designs from partners like Sapphire, XFX, Asus & MSI shouldn't have the issues.
The reference design also seems to be at the power limit, so hopefully cards from their partners will have additional power connections to feed it.
Still, this is a bit sad to see for a brand new card launch and something that's important for AMD.
This is why I think it's always a bad idea to be an early adopter, it reminds me of how Nvidia stung me with the VRAM issue on the 970 which now sits in my Steam Machine (I now have a 980ti in my desktop). Article taken from GamingOnLinux.com.
0 Likes
Some you may have missed, popular articles from the last month:
The comments on this article are closed.
Well I own a gtx 1070 FE which works like a charm. But packaging the debian driver was tricky and now I have the shadow of mordor bug
0 Likes
I have one of those Asus Strix cards, and so does my nephew. My understanding is that any decent mobo already covers for this possibility with components well above the spec, but it's indeed a concern for those interested in overclocking. I'm not, so I should be okay. I did send a technical support query to Asus though, asking if this has been addressed in later revisions of the card and also what this means in practice. It'll be interesting to see what they answer.
0 Likes
It won't cause issues as even older cards draw more from PCIe slot after OC so it won't magically burn your MoBo.
0 Likes
nadrolinux Jul 2, 2016
Eg. some custom GTX 960 had the same 'issue':
http://wccftech.com/amd-rx-480-pcie-power-issue-detailed-overclocking-investigated/
http://wccftech.com/amd-rx-480-pcie-power-issue-detailed-overclocking-investigated/
0 Likes
There is an analysis regarding the power issue there
View video on youtube.com
Last edited by hidekin on 2 July 2016 at 11:45 am UTC
View video on youtube.com
Last edited by hidekin on 2 July 2016 at 11:45 am UTC
0 Likes
TheRiddick Jul 2, 2016
Quoting: GuestAnandtech published an article saying AMD will fix it in their drivers.
I doubt a software patch will fix a fault with a dpm hardware issue on a microprocessor die! Unless they down-clock all 480 card (lol)
0 Likes
TheRiddick Jul 2, 2016
You do realize they compared it to other graphics cards of similar TDP right? And also a few people have had blackouts and broken PCIe slots already right?
Yeah right...
Yeah right...
1 Likes, Who?
Quoting: swickSo, as usual, well thought out and well researched "journalism" (or rather copy and pasting from "tech" blogs) right there liam!I was pretty optimistic at first about the RX 480, then I saw one article about the PCIe problem. I thought it was just some anti-AMD journalist trying to find every single problems he could find, then I saw other articles relating the same stories and not because they were just copy/pasting from other websites, they were actually testing the cards and seeing the same problem. Then I've seen the Hardware.fr article. I know, it's in French, but this site does excellent, in-depth tests of PC hardware parts. I really deeply trust this site, they've always done very accurate tests and benchmark (maybe you can translate it with Google Translate?). And they also said this problem was concerning. PCIe has standards that aren't here just for fun, the manufacturers have to respeect them in order to have everything going on right. If all the PC components were doing what they want with the standards, it would be a hell of a job to build your own PC! Anyway, they said other NVIDIA cards also had the problem indeed, but it wasn't as important as for the RX 480. They ended up saying that, unless you know your mobo really well, they can't advise you to buy this card for now, because it is too risky for the mobo and the other parts connected to it.
Oh, and the comments here: always fucking great.
So, what I'm trying to say is that the common sens here tells us that YOU are probably wrong mister, and every one else seems to be right in that case. And you have to deal with that. I also wanted to have a way better card from AMD, but it didn't happen. I'm disappointed, but you don't see me insulting everyone because of that!
I hope AMD will fix the problem, but until I see the fix (and the benchmarks afterwards), you can be sure I won't buy this card. I really don't want to buy an NVIDIA card, but my card is becoming very old by the day. If the problem isn't fixed, I might go green until the Vega GPU are out.
Last edited by Creak on 2 July 2016 at 1:47 pm UTC
2 Likes, Who?
Quoting: GuestYou've obviously never worked in any sort of industry where fairly complex standards apply. That document may as well be called a guideline instead of a specification. Manufacturers will only stick to it as much as they have to, and there might well be sections entirely broken in the real world. Of course, I'm sure your own thorough investigations into the matter made all that apparent.Swick is right, mobo's should work with the card, if they don't it's a fault of the motherboard and not the card, overclocked cards have been able to go over the 75w specification for ages now. anyone with a motherboard over 50$ shouldn't have any issues with it.
Liam never claimed to do any of the research himself, he simply pointed to an article where people did, or looked to have done a fairly decent look. So stop jumping to such conclusions and posting something that honestly makes you look rather silly.
Last edited by loggfreak on 2 July 2016 at 3:38 pm UTC
0 Likes
Wow, someone actually came in the comments to have a go at me for this even though it's being reported across a large number of sites? Okay then. I deleted his comment as he seemed to edit it out himself.
As has been said, i am pointing out something that is concerning people who do these tests regularly and AMD are confirmed to be looking into it as it is an issue. I did not do the tests myself and never claimed to.
I would do the same for Nvidia, this is not anti-AMD.
If people get annoyed I reference others work, either fund me on Patreon so I can buy the equipment to do it, or bug AMD to get me the card ;)
Last edited by Liam Dawe on 2 July 2016 at 4:47 pm UTC
As has been said, i am pointing out something that is concerning people who do these tests regularly and AMD are confirmed to be looking into it as it is an issue. I did not do the tests myself and never claimed to.
I would do the same for Nvidia, this is not anti-AMD.
If people get annoyed I reference others work, either fund me on Patreon so I can buy the equipment to do it, or bug AMD to get me the card ;)
Last edited by Liam Dawe on 2 July 2016 at 4:47 pm UTC
0 Likes
Consider this a non-issue. Yes the specification is 75W and the card can consume 77W. But a mainboard manufacturer taking only 2 watts of margin into account would be very, very stupid. You can assume safely that any sane mainboard manufacturer designs their board to be capable of delivering more than 75W.
This is mainly bad press for AMD, the amount of situations where it will cause problems will be near zero.
Last edited by dmantione on 2 July 2016 at 5:11 pm UTC
This is mainly bad press for AMD, the amount of situations where it will cause problems will be near zero.
Last edited by dmantione on 2 July 2016 at 5:11 pm UTC
0 Likes
Quoting: dmantioneConsider this a non-issue. Yes the specification is 75W and the card can consume 77W. But a mainboard manufacturer taking only 2 watts of margin into account would be very, very stupid. You can assume safely that any sane mainboard manufacturer designs their board to be capable of delivering more than 75W.Maybe you're right, but from the benchmarks I've seen it wasn't just 2W above the PCIe specs, it was way more than that.
This is mainly bad press for AMD, the amount of situations where it will cause problems will be near zero.
The PCIe specs says the maximum intensity the card should get from the mobo is 5.5A but, in Battlefield 4, it's up to 6.92A (that's 25% above the specs without overclocking anything!). Other graphics cards can do it simply because they have more than just a 6-pin connector. AMD should have at least an 8-pin connector by default on their RX 480.
Of course it is good if the mobo can handle it, it means that the mobo can behave correctly if you overclock some parts on it. But that's not how standards work! By default, any component should respect the specs. If it is indeed only 2W... then I promise I won't make such a fuss about it, but it doesn't seem so (I mean just read the articles, they don't even want to try overclocking because they're afraid to damage the other components on their test machines).
A good example of manufacturers not respecting the power specs and leading to dangerous products is the bad USB-C cables. That lead to burn your battery, and clearly it's not up to your phone to handle bad USB-C cables.
I really want AMD to do something about it, I really do, because I really want to buy an AMD card. But until a fix is found (whether it's hardware or software), I think it's stupid to buy an RX 480 card.
Last edited by Creak on 2 July 2016 at 5:59 pm UTC
0 Likes
I must admit, I've read several articles, but not the one linked by GoL. So I just read it and I agree that the conclusions are far less catastrophic than what I've been used to read. It says that the power consumption isn't that high compared to other cards, and good custom cards that will be sold later should fix the problem by adding a 8-pin connector.
That being said, even in regards to this reading, I'm still disappointed by AMD who advertised a low-power consumption GPU and apparently tried to squeeze the card into this definition by using a 6-pin connector and being just above the limits of the PCIe specs. That is bad PR. It gives the impression that the GPU, although a brand new architecture, has already reached its limits. And it also gives bad press to the cards. It would have been better to have a product a bit less on the edge, at least for the first revisions. Now a lot of users won't even look twice at those cards because of what happened.
Because I really prefer AMD over NVIDIA for several personal reasons, I will continue to follow closely this topic and hope for a brighter tomorrow with the custom cards that seems to be on the way. But if the problems aren't fix by then, I'll be extremely disappointed by AMD.
Last edited by Creak on 2 July 2016 at 7:08 pm UTC
That being said, even in regards to this reading, I'm still disappointed by AMD who advertised a low-power consumption GPU and apparently tried to squeeze the card into this definition by using a 6-pin connector and being just above the limits of the PCIe specs. That is bad PR. It gives the impression that the GPU, although a brand new architecture, has already reached its limits. And it also gives bad press to the cards. It would have been better to have a product a bit less on the edge, at least for the first revisions. Now a lot of users won't even look twice at those cards because of what happened.
Because I really prefer AMD over NVIDIA for several personal reasons, I will continue to follow closely this topic and hope for a brighter tomorrow with the custom cards that seems to be on the way. But if the problems aren't fix by then, I'll be extremely disappointed by AMD.
Last edited by Creak on 2 July 2016 at 7:08 pm UTC
1 Likes, Who?
Another point of view for this is the adoredtv video about it. He points out that even though the average Watts is a little higher than other cards, many cards have significantly higher spikes in consumption with a lower average usage and that this is a common problem throughout the industry. Also, these high averages only occur with resolution, game, and quality settings that you would never run in real life because they're basically a slide show. Additionally a prior video notes that the best efficiency of power to performance on the card is when you set it to -20% power consumption. The new wattman software has fairly fine gained control of the power consumption at various temperatures and clock speeds so under clocking it just in these torture tests in software to prevent the problem should be easy with minimal performance drops in normal operation.
0 Likes
The funny part is already cards having the same issue, and no one make a big deal.
If people that made so many articles and afraid to damage their hardware they just don' t need to do the test on their hardware usually places that make this kind of measures or testing, have plenty of hardware to reproduce the error or try to damage their hardware in some situations.
So people are making a big deal where is not. Let the places that make review do their test, then wait for the company to respond to that experiments found it.
Anyway the only things is that the graphics cards are demanding more power from the pci express, than usually. That is the only problem if this situation make failure hardware then we need to worry, they need do a better research and for that they need time.
About Creak said Amd said it will work on 6 pin connectors, and it does, isn't it? Remember when you make an specification for anything you can make tolerance, between some ratio that make the hardware works without problem, even when you calculate something you need to do that or to the found an error.
The main problem here is that articles are making a huge deal with the topic then when you read, the topic is already misleading. Even in GOL "The AMD RX 480 seems to have power issues you may want to be aware of". The good things is explain and give some good advice as others company been in the past.
If people that made so many articles and afraid to damage their hardware they just don' t need to do the test on their hardware usually places that make this kind of measures or testing, have plenty of hardware to reproduce the error or try to damage their hardware in some situations.
So people are making a big deal where is not. Let the places that make review do their test, then wait for the company to respond to that experiments found it.
Anyway the only things is that the graphics cards are demanding more power from the pci express, than usually. That is the only problem if this situation make failure hardware then we need to worry, they need do a better research and for that they need time.
About Creak said Amd said it will work on 6 pin connectors, and it does, isn't it? Remember when you make an specification for anything you can make tolerance, between some ratio that make the hardware works without problem, even when you calculate something you need to do that or to the found an error.
The main problem here is that articles are making a huge deal with the topic then when you read, the topic is already misleading. Even in GOL "The AMD RX 480 seems to have power issues you may want to be aware of". The good things is explain and give some good advice as others company been in the past.
0 Likes
Quoting: TheRiddickI doubt a software patch will fix a fault with a dpm hardware issue on a microprocessor die! Unless they down-clock all 480 card (lol)
Remember those on-chip microcode images we load at power up that some people complain about ? If a software update is not sufficient and the DPM logic actually needs to change that just means an updated microcode image...
2 Likes, Who?
Hello there, all over the place I read about powergate, while other reviews completely skip this aspect. While this is not good for AMD for PR reasons, all their efforts towards Linux are not forgotten. I would love to buy myself a gaming PC (never had, just reading gaming articles with longing), but I would love to have zen onboard too. I may wait for the combo. At work hoeever we use Linux workstations for cfd and simulations and we use quadro cards, but the administration is not simple with different driver versions for old, very old and new cards. I will suggest that once amdgpu supports well recent firepro cards we buy one of them and test it.
Need to support a manufacturer with good Linux stance...
Need to support a manufacturer with good Linux stance...
1 Likes, Who?
Quoting: bridgmanQuoting: TheRiddickI doubt a software patch will fix a fault with a dpm hardware issue on a microprocessor die! Unless they down-clock all 480 card (lol)
Remember those on-chip microcode images we load at power up that some people complain about ? If a software update is not sufficient and the DPM logic actually needs to change that just means an updated microcode image...
Is that AMD-Bridgman ?
0 Likes
Quoting: raonlinuxThe main problem here is that articles are making a huge deal with the topic then when you read, the topic is already misleading. Even in GOL "The AMD RX 480 seems to have power issues you may want to be aware of". The good things is explain and give some good advice as others company been in the past.The article title is directly related to what is being reported by many sites. I said "seems" for a very good reason as it shows ambiguity/speculation since I didn't do the testing myself, but if multiple people are claiming it and AMD is looking into it that should tell you there is some issue here.
Quoting: bridgmanNice to see you here :)Quoting: TheRiddickI doubt a software patch will fix a fault with a dpm hardware issue on a microprocessor die! Unless they down-clock all 480 card (lol)
Remember those on-chip microcode images we load at power up that some people complain about ? If a software update is not sufficient and the DPM logic actually needs to change that just means an updated microcode image...
0 Likes
Quoting: pete910Is that AMD-Bridgman ?
Yessir...
1 Likes, Who?
Patreon
PayPal
See more from me