AMD once again violating power specifications? (AMD RX-480)

[Trixanity]

2 hours ago, Trixanity said:

PCPer has tested the 960 Strix on user request and did not find any problems with their sample. Whether it's the exact same model and revision is unclear but what is clear is that they did not find issue with it.

 

PS. I'm posting this for one reason only: full disclosure. I do wish the thread would die but it probably won't.

I do wish AMD would come forth with a solution soon so we can get this over with. Rumors said today but we haven't heard anything yet, so I guess it'll be next week considering it's Friday. AMD can't afford to be quiet for much longer though; the backlash is massive.

 

3 minutes ago, LAwLz said:

Hey @Trixanity, look at all the people who are now back with the whole " Nvidia is doing the same thing! Look at the 960 Ti Strix!" again.

See why I was so worried about your joke post now?

 

Personally I give up. AMD fanboys are too retarded to be educated. Look at all of them taking Jayz' baseless claims as absolute facts just because they agree with their predefined conclusion. Look at them agreeing with AdoredTV which is a collosal AMD fanboy who doesn't know what he is talking about (again, using the same picture you used in your joke post).

 

 

I wish I could enforce a rule that if your IQ is below the average PCIe slot power draw of the RX 480 then you wouldn't be allowed to talk about it...

Let me point you to an earlier quote. I've already made a statement on that. It was meant to please you. I guess it did not or it was overlooked.

Let's face it: all fanboys are... less than ideal to put it mildly. And also, pretty much everyone values sources that confirm their own ideas and opinions. And that goes for everything. That's why some watch CNN and others watch Fox News. They want to hear their own opinion repeated back to them. They don't want to be challenged. I can't help you fix that - no one can.

[LAwLz]

2 minutes ago, laminutederire said:

I was just trying to explore an explanation which was submitted, that doesn't mean you can be insulting.

Searching for the reason it does that isn't what we should talk about here?

I was just saying that the idea that it appears in obviously ridicule usage of the card could point out to a reason why AMD has un professionally overlooked it and may point towards a direction to understand why it behaves like that.

Oh sorry my post wasn't directed at you.

[zMeul]

1 hour ago, Jahramika said:

LOL dude he list 3 cards you shill and this is the AMDs first thats been reported

the 1st?! have you heard of HD6990 that didn't even had PCIe logo on the box because AMD didn't even bothered to send it to PCI-SIG for certification?

 

http://www.anandtech.com/show/4209/amds-radeon-hd-6990-the-new-single-card-king/5

[laminutederire]

3 minutes ago, LAwLz said:

Oh sorry my post wasn't directed at you.

No worries, I destroyed myself my vram path though.

What could overdraw power due to 4K resolution ? 

Maybe shader units? 

[TheRandomness]

Why have people been arguing about this for 35 pages?

[laminutederire]

6 minutes ago, zMeul said:

the 1st?! have you heard of HD6990 that didn't even had PCIe logo on the box because AMD didn't even bothered to send it to PCI-SIG for certification?

 

http://www.anandtech.com/show/4209/amds-radeon-hd-6990-the-new-single-card-king/5

That's because AMD has balls you know

 

[zMeul]

1 minute ago, laminutederire said:

That's because AMD has balls you know

yes Harry, they are waazards

[laminutederire]

1 minute ago, zMeul said:

yes Harry, they are waazards

Don't they have a huge center in Texas? That'd explain their cowboy-ish look towards standards ! 

[gilang01]

2 hours ago, gilang01 said:

JagatReview's overclocker is testing the power gate with a 1425 mHz rx480 (there's a thread on reddit for this particular gpu) and a H81 mobo. I'll update the results as soon as it's out

OK here's an update for my previous post. 

JagatReview tested their overvolted 1150mV, overclocked rx480 @1425 on a cheap H81 and so far no damage has been observed. They stressed tested it for 2 hour (still running) with Uni engine Heaven. 

Will post another update if something comes up

[leadeater]

4 hours ago, LAwLz said:

I wish I could enforce a rule that if your IQ is below the average PCIe slot power draw of the RX 480 then you wouldn't be allowed to talk about it...

HAHA, I'd make a joke about the internet in general.... but I won't .

 

Also OMG 35 pages, better all be worth reading..... but this post is also contributing to that problem so who am I to criticize. 

[SukaroBlue]

Going to play devil's advocate here. maybe there was manufacture defect in the batch that shipped and that's what causing the problem. AMD would still be at fault for not testing the batch though.

[stconquest]

10 minutes ago, SukaroBlue said:

Going to play devil's advocate here. maybe there was manufacture defect in the batch that shipped and that's what causing the problem. AMD would still be at fault for not testing the batch though.

R e a d   t h e   t h r e a d.

[Pendragon]

 

[Wolther]

This isn't something to worry about.. Atleast to change your mind from buying the card or not. Many other cards exceed this power limit. From what I believe, there is lots of headroom from motherboard manufacturers. A 750 TI uses like 70-74? watts from the connector, imagine anyone OCing on that card.. Wow they all must've lost their motherboards then.. Nope. they didn't

 

Although I do agree that they should've just used something better than 6 pin.. But I have doubt it'll affect your MB  

[stconquest]

26 minutes ago, Pendragon said:

 

That has been posted here already.

[Briggsy]

29 minutes ago, Pendragon said:

 

More recent tests, like the one done by PCper using witcher and tomb raider, were showing 80-85 watts being pulled from the slot 12v rail alone at stock, with 100+ watt averages overclocked. And, these were 1080p tests. It's not just the wattage, but also the fact the 480 Amperage goes way over spec and causes vdroop, and the motherboard traces in turn have to dissipate 3-4 watts of heat.

 

The problem with adoredTV is that he sounds like hes preaching the gospel truth if you agree with him (confirmation bias). but, if you are trying to be objective, his arguments sound like a house of cards. In this case, his reasoning is just that, sadly.

[stconquest]

13 minutes ago, Briggsy said:

More recent tests, like the one done by PCper using witcher and tomb raider, were showing 80-85 watts being pulled from the slot 12v rail alone at stock, with 100+ watt averages overclocked. And, these were 1080p tests.

 

The problem with adoredTV is that he sounds like hes preaching the gospel truth if you agree with him (confirmation bias). but, if you are trying to be objective, his arguments sound like a house of cards. In this case, his reasoning is just that, sadly.

Watching this caused me to solidify a belief that this will end with AMD keeping their mouth shut and having all this swept under the rug.

 

Best line in the video:  Which one is worse, the 960 Strix with it's 225W peaks or the RX 480 with it's average well above 75w (12v rail + 3.3v rail)? 

 

Answer:  Both are fucking bad.  To have such readily available safeguards to stay well within spec, both these manufacturers decided to rely on the "overbuilt" motherboards.

[Briggsy]

34 minutes ago, stconquest said:

Watching this caused me to solidify a belief that this will end with AMD keeping their mouth shut and having all this swept under the rug.

 

Best line in the video:  Which one is worse, the 960 Strix with it's 225W peaks or the RX 480 with it's average well above 75w (12v rail + 3.3v rail)? 

 

Answer:  Both are fucking bad.  To have such readily available safeguards to stay well within spec, both these manufacturers decided to rely on the "overbuilt" motherboards.

Yeah, it is something I've always wondered about. Why are cards without any pci-e connectors given the ability to overclock to begin with if they are already near 75watts. I suppose the answer comes down to 6 parts in this discussion about the 480 (at least).

 

1 - those 75 watt cards don't break spec out of the box, even with gpu boost (Actually gpu boost brings the cards up to 75w average before stopping - in the case of Nvidia). The key point being average power draw. spikes are meaningless because that's what capacitors are for.

 

2 - In order to overclock those 75w cards, you need to download 3rd party tools, which come with their own disclaimer on top of the GPU vendors disclaimer about overclocking damage and how you overclock at your own risk.

 

3 - The 480 breaks 75w slot spec at stock clocks

 

4 - The 480 draws extra power from the slot after reaching 150w power draw, instead of the pci-e connector. This can't be accidental.

 

5 - AMD provide extensive overclocking tools within radeon settings, launched alongside the reference 480. Despite the disclaimer you have to agree to in Wattman, they are actively encouraging 480 owners to overclock their reference 480's.

 

6 - Modern PSU's that have pci-e connectors all come standard as 6+2pin. It's been this way for probably 5 years at least. Even then, some graphic cards ship with a single 6-pin to 8-pin adapter, because the 6-pin is capable of handling double the power draw. At best, AMD should have made the 480 with an 8-pin connector, shipped the cards with a 6-pin to 8-pin adapter, and designed the 480 to draw excess power from the auxiliary connector instead of the slot.

[stconquest]

8 minutes ago, Briggsy said:

Yeah, it is something I've always wondered about. Why are cards without any pci-e connectors given the ability to overclock to begin with if they are already near 75watts. I suppose the answer comes down to 6 parts in this discussion about the 480 (at least).

 

1 - those 75 watt cards don't break spec out of the box, even with gpu boost (Actually gpu boost brings the cards up to 75w average before stopping - in the case of Nvidia). The key point being average power draw. spikes are meaningless because that's what capacitors are for.

 

2 - In order to overclock those 75w cards, you need to download 3rd party tools, which come with their own disclaimer on top of the GPU vendors disclaimer about overclocking damage and how you overclock at your own risk.

 

3 - The 480 breaks 75 slot spec at stock clocks

 

4 - The 480 draws extra power from the slot after reaching 150w power draw, instead of the pci-e connector. This can't be accidental.

 

5 - AMD provide extensive overclocking tools within radeon settings, launched alongside the reference 480. Despite the disclaimer you have to agree to in Wattman, they are actively encouraging 480 owners to overclock their reference 480's.

 

6 - Modern PSU's that have pci-e connectors all come standard as 6+2pin. It's been this way for probably 5 years at least. Even then, some graphic cards ship with a single 6-pin to 8-pin adapter, because the 6-pin is capable of handling double the power draw. At best, AMD should have made the 480 with an 8-pin connector, shipped the cards with a 6-pin to 8-pin adapter, and designed the 480 to draw excess power from the auxiliary connector instead of the slot.

Now that is exactly where we are.  Why would you split the power draw between the slot and cable 50/50?  AMD is not new at this.  They know this card will go beyond the 150W mark.  They know the specs on the 6 pin cable are more robust on a broad scale, and can handle more power than the mobo. 

 

Personally, I have such a hard time understanding why.  I know there is a reason, just have no idea where to start looking.  It really can't be incompetence, there is another reason they went this route, there has to be.

[MoonSpot]

PcPer latest youtube upload.  Haven't watched it yet, will add comment once I have.

 

:EDIT:

The TL;DR/W is:

• Power draw overage for extended periods through PCI-e confirmed, tis not just spikes.
• AMD working on a patch/fix, no answer as to what since we're barely 48hrs in after release.
• Other cards which people suspect were also pulling more power through mobo almost certainly were not.
• Ryan has issues with pointing a stylus away from the screen when trying for touch functionality.

 

Full story link : https://www.pcper.com/reviews/Graphics-Cards/Power-Consumption-Concerns-Radeon-RX-480

 

:EDIT 2: Ha!  Sorry, didn't realize how late to the party I was here.  Have been drinking.

[laminutederire]

1 hour ago, MoonSpot said:

PcPer latest youtube upload.  Haven't watched it yet, will add comment once I have.

 

They have the same results Tomshardware has, but they seem much more professional, since we actually learn things thanks to their video.

[Syntaxvgm]

drawing over 75w on the board? Yikes, wtf. 

[HKZeroFive]

@Majestic, might be worth adding this to the OP. It's a statement from AMD themselves:

Quote

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).

So it appears it can be fixed on a software level? Going to remain very skeptical until it's confirmed.

[Trixanity]

Official statement from AMD:

Quote

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).

I called it on the software providing a fix instead of a recall. Although I'll be fair and say it might not be a 'perfect' fix meaning there might be caveats. 

 

I'll re-iterate what I've stated before: these cards are running at too high a voltage. It's actually hurting their performance by boosting temperature which lowers performance to keep everything in check. The cards run at 1.3V default I believe and 1.1V is easy to do and some reports of even lower voltage with fine-tuning. I'm not sure if they'll try to lower the voltage through software as it could be risky for inexperienced users (despite some evidence suggesting they'll underclock to certain points almost guaranteed) who can't circumvent a bad change causing instability. Another option might be software changes to how power is requested which could fix the pcie bus problem (if this is possible through software), lower power targets is also an option. Underclocking is also an option but not really ideal as they've sold the cards with a frequency advertised and by extension a performance number.

 

Edit: they mention the speed of the memory interface but I think it's just a red herring so to speak. I don't think the problem is with the memory and its power draw. I think the memory is kinda spoken for if you will; my point being that the implementation is fairly standard and therefore without curve balls. Well, unless AMD knows something I don't which is very likely (that would be an understatement). At best, it's the memory drawing a lot of power from the bus (it uses what? 40W on the 8 GB model?) and in conjunction with the graphics chip it ends up being too much but they would have known that up front when opting for that memory solution.

[stconquest]

That's awesome.  I wonder what they will change.  Will clocking down the RAM be enough...