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

[Prysin]

5 minutes ago, Majestic said:

This isn't the first time you apologetically show up to defend the 295X, but fine i'll remove the last scentence if you will edit it out aswell.

However you were accusing me of cherrypicking and using outdated standards as a way to make a point, which makes you apologetic to tossing out specs and standards to rationalize your purchase. Which is very annoying.

why should i edit it out? are you not going to stand by what you said?

 

Every single insult or profanity i say, i leave in my posts. You know why? Because i do not care if my actions get repercursions. I am enough of a man to stand by my word.

 

As for the 295x2. I am not "apologetic". i have nothing to apologize for. I knew well in advance of buying it how much power it would draw, how much the wires could handle, how much my PSU could handle. Which PSUs would be safe to use with it.

 

You however are just going on a blind rampage, about specifications that were invalidated in march 2007. About things that can with basic mathematics be proven save. Thing is. you are trying to argue with someone who work with wiring for a living. You were on the losing side before you even started to argue.

 

I am not apologizing to anyone for AMDs products. They can do that themselves if they want to.

I am simply here to correct your bullshit before people  think it is actually true. Unfortunatly for you, i got evidence on my side.

[handymanshandle]

Wouldn't it only be a problem on the reference card and that aftermarket cards could fix this voltage issue?

[Prysin]

4 minutes ago, wcreek said:

Wouldn't it only be a problem on the reference card and that aftermarket cards could fix this voltage issue?

yes. However there are thousands of reference models out there by now. A recall would be massively damaging to the brand name. SO the best way to fix it is with a sneaky vBIOS update.

[handymanshandle]

1 minute ago, Prysin said:

yes. However there are thousands of reference models out there by now. A recall would be massively damaging to the brand name. SO the best way to fix it is with a sneaky vBIOS update.

So it's not a driver limitation but rather what AMD did with the BIOS on the card? Because what I've gathered is that the software voltage regulation makes OCs hard or impossible.

[Sauron]

Tom's hardware recently posted a grossly inaccurate article on the 480's pricing, so I would take this with a grain of salt.

[Majestic]

13 minutes ago, Prysin said:

As for the 295x2. I am not "apologetic". i have nothing to apologize for. I knew well in advance of buying it how much power it would draw, how much the wires could handle, how much my PSU could handle. Which PSUs would be safe to use with it.

 

You however are just going on a blind rampage, about specifications that were invalidated in march 2007. About things that can with basic mathematics be proven save. Thing is. you are trying to argue with someone who work with wiring for a living. You were on the losing side before you even started to argue.

 

I am not apologizing to anyone for AMDs products. They can do that themselves if they want to.

I am simply here to correct your bullshit before people  think it is actually true. Unfortunatly for you, i got evidence on my side.

You are apologizing for AMD's liberate usage of power consumption per PEG connector. All cards managed to say within the 150W per connector and AMD decides to throw a spanner in the works and doesn't simply put 3 connectors on the board to save board space and a few bucks. How is this consumer-friendly, and how are you not apologizing bad , anti-consumer, behaviour by condoning it or writing it off. 

 

You might know the insides of a PSU, and what is safe usage. But without simple guidelines, everyone who is building PC's has the same knowledge barrier before they are able to build it. Because some manuf. thought it'd be easier to ignore simple standards in terms of power consumption per connection and just wing it.

 

Also, I'm again not arguing about it phsyically working or not, I'm an Mechatronics Engineer aswell, if you insist on making this personal. I have only ever argued the principle of having regulation standards (enfored or not), and it being much more consumer friendly. You sir, are very, very anti-consumerist and smug about it too.

 

No you're right, I do stand by my earlier remark. 

[Prysin]

1 minute ago, wcreek said:

So it's not a driver limitation but rather what AMD did with the BIOS on the card? Because what I've gathered is that the software voltage regulation makes OCs hard or impossible.

it may also be solved through drivers. But atm the issue is, well allegedly so, that the RX 480 draws more power from the PCIe port then it was designed to withstand. This is a serious issue and should not be taken lightly (happened with Nvidias MAXWELL GPUs too).

It can be solved through three means

 

A) A product recall and retrofitting the PCB with a 8-pin connector

B) specify through vBIOS that the card shall draw more power from the 6 pin and less from the PCIe. The 6 pin is rated for minimum 75w, however it's theoretical maximum is actually 150w or so.

C) nerf the cards clock speeds through drivers/vBIOS.

[GzeroD]

Let's all stay on the topic

[RagnarokDel]

4 hours ago, Majestic said:

Because it peaks 155W over a bus designed for 75W. Meaning it will stress the mainboard momentarily for double it was designed for.

They note it won't catch fire, but it will cause perhiperal issues like sound or add-in cards.

 

But who knows when someone is using a cheap board and is overclocking it. Meaning it will average 100W over something designed for 75W. It could cause damage.

No it doesnt. 155w - 75 from the 6 pin= 80W, 5 more watts.

[Majestic]

9 minutes ago, Prysin said:

 (happened with Nvidias MAXWELL GPUs too).

Source?

 

EDIT: 

http://www.tomshardware.com/reviews/nvidia-geforce-gtx-980-970-maxwell,3941-12.html

 

The 970 and 980 sure didn't.

[AresKrieger]

I feel like the 480 will not be the card that saves amd, but it will be the subject of flame wars for months to come. Honestly though the hype was too strong for this thing to succeed to expectation even without issues popping up, either way this could be entertaining for a while.

[handymanshandle]

2 minutes ago, Prysin said:

it may also be solved through drivers. But atm the issue is, well allegedly so, that the RX 480 draws more power from the PCIe port then it was designed to withstand. This is a serious issue and should not be taken lightly (happened with Nvidias MAXWELL GPUs too).

It can be solved through three means

 

A) A product recall and retrofitting the PCB with a 8-pin connector

B) specify through vBIOS that the card shall draw more power from the 6 pin and less from the PCIe. The 6 pin is rated for minimum 75w, however it's theoretical maximum is actually 150w or so.

C) nerf the cards clock speeds through drivers/vBIOS.

Suppose those are options.

 

[Majestic]

2 minutes ago, RagnarokDel said:

No it doesnt. 155w - 75 from the 6 pin= 80W, 5 more watts.

guess ill have to post it again

 

 

That 155W WAS the mainboard powerspike.

[Dackzy]

come on this problem is not that big, they just need to make a vBIOS update, so the card will draw more power from the 6 pin than it does right now and that would not be a problem for the 6 pin, as prysin said the 6 pin can theoretically go up to around the 150w mark.

[Majestic]

4 minutes ago, Dackzy said:

come on this problem is not that big, they just need to make a vBIOS update, so the card will draw more power from the 6 pin than it does right now and that would not be a problem for the 6 pin, as prysin said the 6 pin can theoretically go up to around the 150w mark.

As I said earlier, the 295X2 had the benefit of being a niche product that will attract a....niche audience like Prysin. This will be a bulk unit, that will attract the budget buyer. Meaning it won't necessarily always run on overdesigned power supplies. 

 

I think this is a big deal, they need to throttle the card, not relocate the problem.

[Sparviero]

A budget card going in budget systems designed to overmatch the pcie power sounds like a horrible combo. AMD surely knows there will be systems with iffy power issues.  I doubt there was anything nefarious planned. Maybe I'm wrong though. *shrug*

 

The last time there was a discussion of the 8 vs 6 pin power during the 1080 reference board launch didn't someone say one of the pins does nothing and the other is an extra ground?

 

thanks for clarification. 

[GzeroD]

I find it odd how the Anandtech review has the total system power consumption lower or about equal to what toms has for only the "VGA card total"

[Trixanity]

2 minutes ago, Sparviero said:

A budget card going in budget systems designed to overmatch the pcie power sounds like a horrible combo. AMD surely knows there will be systems with iffy power issues.  I doubt there was anything nefarious planned. Maybe I'm wrong though. *shrug*

 

The last time there was a discussion of the 8 vs 6 pin power during the 1080 reference board launch didn't someone say one of the pins does nothing and the other is an extra ground?

 

thanks for clarification. 

Somewhat correct. One pin is ground, the other is a 'sense' pin that basically signals "I'm an 8-pin". However, it would also seem one of the pins (a 12V pin) on the 6-pin aren't active but are on the 8-pin. So there's that.

[Majestic]

4 minutes ago, GzeroD said:

I find it odd how the Anandtech review has the total system power consumption lower or about equal to what toms has for only the "VGA card total"

Because system power is less accurate. Different cards can have different CPU loads (driver efficiency), meaning they can obfuscate the amount the card draws. It can also mean they ran a different scenario than Tom's hardware. Try running either Crysis 3 or Unigine Heaven. My system went up from 220W to 320W running Crysis 3.

 

It seems Tom's hardware uses a benchmark that specifically loads the board to it's full capacity (memory and GPU). They also explain as much in the review btw.

[Dackzy]

4 minutes ago, Majestic said:

As I said earlier, the 295X2 had the benefit of being a niche product that will attract a....niche audience like Prysin. This will be a bulk unit, that will attract the budget buyer. Meaning it won't necessarily always run on overdesigned power supplies. 

 

I think this is a big deal, they need to throttle the card, not relocate the problem.

it does not need a overdesigned PSU in any way, it just needs to not be a cheap china PSU, if it is a okay quality one then you are fine and we are not even talking high quality.

[maulemall]

I just want to thank toms hardware because if it wasn't for them we wouldn't know the sky was falling.

Please ..Can we get this thread pinned so we can find it first thing everyday ??

That evil AMD can't get away with this.

[Briggsy]

I think this is the point when the topic of paper launches becomes relevant again in the discussion.

 

I've always been in favour of benchmarks/reviews preceding a launch by a few days at least. I believe it gives time for problems to be weeded out and removed from the public eye beforehand. I'm sure there are still people that are cool with benchmarks coming out on the launch day, but they probably haven't bought a 480 before finding out their cheap motherboard might melt.

[mariushm]

It's not quite as easy, if the pcb is constructed as Tom's Hardware says so.

According to TH, the VRM is a six phase design and 3 phases are wired to the pci-e slot, and 3 wires are wired to pci-e connector so in theory the board draws evenly from both sources.

The 150w spikes are just that, spikes, for very small periods of time like 1ms or something like that. 150w at 12v is only about 12A - for a ms or something like that, there's not enough time for pcb traces or the pci-e slot pins to overheat from the excessive current.

 

It's true however that onboard devices like audio cards could "pick up" these sudden pulses of current and reflect them as audible sound.

 

To the best of my knowledge, the standard says the pci-e slot is supposed to deliver only 60 watts on 12v and the rest of 15 watts on 3.3v, not sure where that 75 watts figure comes from. 

But, the power consumption is still a bit too much, my guess is that it's due to raising the base frequency at the last minute.. as far as I know the base memory was supposed to be 1080 Mhz but the cards showed up at 1120 Mhz , maybe the extra frequency got the power consumption over the 150 watts value they initially designed the card for.

 

[SageOfSpice]

I'm not really going to trust this until I see enough people test it.

 

If this is real life though, that's a pretty glaring issue. I find it hard to believe that would get past product testing.

Then again, I don't exactly work at AMD so I don't know their policies.

 

I'm really hoping deep inside this isn't just a big flop, though. 

This was a pretty good opportunity, it'd be a shame for it to be squandered. 

[Sauron]

21 minutes ago, AresKrieger said:

I feel like the 480 will not be the card that saves amd, but it will be the subject of flame wars for months to come. Honestly though the hype was too strong for this thing to succeed to expectation even without issues popping up, either way this could be entertaining for a while.

Consider that those who are discussing it so much are within the enthusiast community, which is not necessarily the majority of customers. At the end of the day for a product to sell well you need good performance for the buck and good marketing. AMD has the first but it has always lacked the latter. That doesn't mean this will be a failure for them, but it certainly won't place them ahead in the market. If they ever get to that point it will be a slow process, but it could start now.