The 480: power consumption, PCI-E powerdraw

[IEC]

The miner burned his board at the +12V ATX power connector. My guess is resistance at the connection point was higher than it should have been, leading to the melt once significant power was drawn through it. It's rather telling he did not burn the PCI-e 1x slot even though he was drawing significantly over the 25W typically associated with 1x slots.

I would be very surprised if people are actually burning out PCI-e slots, given that the first point of failure in most systems would likely be at that very same +12V connection to the motherboard. Which in itself is still difficult to manage except in pathological cases such as mining without powered risers or motherboard auxiliary +12V.

 

[Abwx]

960's aren't burning up boards, 480's are. Fact. Making up your own PCIe standards won't change that.

Provide us with an exemple of your "facts", and not some kind of shills like the two brazilians that you linked and who i caught pant down because they thought that Furmark would be more power consuming than regular games, so where is your exemple..?.

 

[2is]

Provide us with an exemple of your "facts", and not some kind of shills like the two brazilians that you linked and who i caught pant down because they thought that Furmark would be more power consuming than regular games, so where is your exemple..?.

"provide me an example, just not one that proves me wrong" lmao.

It doesn't matter how much furmark consumes compared to games. They proved that the card draws WAY too much power from the PCIe slot, this was shown in furmark as well as other tests they ran. Every sinlge test consumed more power then it should. So your reasons for invalidating that video are complete nonsense.

Facts are facts, evidence is evidence, but go ahead and continue making up your own reality, complete with your own PCIe standards.

https://www.youtube.com/watch?v=nKcHR1qW3w4&feature=youtu.be

 

[el etro]

Provide us with an exemple of your "facts", and not some kind of shills like the two brazilians that you linked and who i caught pant down because they thought that Furmark would be more power consuming than regular games, so where is your exemple..?.

I know this guy, he's really an arse. Very PRO Nvidia(at AlienBabelTech levels) and also Galax close partner, Galax GTX Kingpin edition equivalent comes with his signature. He's one of the most known OC'ers on the world.

 

[Bacon1]

"provide me an example, just not one that proves me wrong" lmao.

It doesn't matter how much furmark consumes compared to games. They proved that the card draws WAY too much power from the PCIe slot, this was shown in furmark as well as other tests they ran. Every sinlge test consumed more power then it should. So your reasons for invalidating that video are complete nonsense.

Facts are facts, evidence is evidence, but go ahead and continue making up your own reality, complete with your own PCIe standards.

https://www.youtube.com/watch?v=nKcHR1qW3w4&feature=youtu.be

He was asking for proof that boards were being damaged, not that it is possible to pull more wattage. Multiple other videos have done so in the past.

 

[Abwx]

It doesn't matter how much furmark consumes compared to games. They proved that the card draws WAY too much power from the PCIe slot, this was shown in furmark as well as other tests they ran.

Yes and Furmark use 120W at the card level, while they are stating that the whole 120W are supplied through the PCIe connector, wich is why they are obviously lying, in Furmark the PCI connector will hence provide much less than 120W, the only way to get this is to remove the 6 pin cable and do a soldering on the card to connect the PCIe to the 6 pin routing, what they likely did.

150W delta at the main, wich means barely 135W for the card, you understand better why i call them liars and shills..?.
That s indeed the only thing they managed to prove..

So here your poor "fact", shills doctoring a test and yourself parroting their lies despite all the technical evidences since i already posted this AT page...

I know this guy, he's really an arse. Very PRO Nvidia(at AlienBabelTech levels) and also Galax close partner, Galax GTX Kingpin edition equivalent comes with his signature. He's one of the most known OC'ers on the world.

Thanks to AT measurements their despicable viral marketing is debunked, and in my opinion they didnt do this for free...

 

[Abwx]

He was asking for proof that boards were being damaged, not that it is possible to pull more wattage. Multiple other videos have done so in the past.

Of course he cant provide anything, hence why he changed the goal post,
alas it just exposed that he s using viral marketing videos as a mean to badmouth AMD, anyone who would had read a few review would had noticed the Furmarks numbers being well below games power comsumption..

 

[Bacon1]

Radeon RX 480 vs Cheap/Budget Motherboards tested here - https://www.youtube.com/watch?v=LxiZqQVXuzA

 

[SimianR]

Do we actually have any examples of the 480 killing someone's motherboard when not in an extreme situation like the guy mining with 3 cards and non-powered risers?

 

[FFFF]

I think a class action lawsuit is in order here. AMD engineering/QC teams are a joke. :thumbsdown:

 

[Abwx]

Do we actually have any examples of the 480 killing someone's motherboard when not in an extreme situation like the guy mining with 3 cards and non-powered risers?

There s none but we have someone here who say otherwise, and who will surely be very happy to provide us the "facts" he s still holding as an exclusivity :

960's aren't burning up boards, 480's are. Fact. Making up your own PCIe standards won't change that.

 

[Doom2pro]

8-pin connectors are rated for 150W...

Keep searching!

One of those 8 pins is OPTIONAL, without it, it's 125W.... Not all PSUs utilize the 8th pin (which is a reserved Sense pin), and not all cards utilize it either.

 

[Doom2pro]

I think a class action lawsuit is in order here. AMD engineering/QC teams are a joke. :thumbsdown:

Far from it, they have created a platform that is competing with a VASTLY better funded nVidia...

You create a Graphics startup at 1/10th the budget of nVidia and release products this competitive.

Good luck.

 

[FFFF]

Far from it, they have created a platform that is competing with a VASTLY better funded nVidia...

You create a Graphics startup at 1/10th the budget of nVidia and release products this competitive.

Good luck.

They got in this bad shape through their own failures mostly. Consumers don't care about the financial situation of a company, they just want their product properly tested and engineered so it works properly.

 

[Sweepr]

150W delta at the main, wich means barely 135W for the card, you understand better why i call them liars and shills..?.

210W (average gaming minus idle) in ComputerBase's review at 'Max':

The Maximized Power Target allows the power consumption of the Radeon RX 480 to rise significantly. 267 watts required for the graphics card in consistently applied 1,266 MHz, which additional 42 watts means.

Nowhere near 135W.

 

[mutantmagnet]

Radeon RX 480 vs Cheap/Budget Motherboards tested here - https://www.youtube.com/watch?v=LxiZqQVXuzA

Cool another data point. I didn't realize the ScienceStudio budget board was AM2. That's really old and it conforms to the idea that later versions of PCIE are built better than the 1.0 spec.

This guy was using an AM3 board that is hardly modern but still a more likely pick up for budget oriented users today. The reference can be handled better with today's motherboards even the cheaper ones. That said it's clear running over spec can cause issues if a board isn't designed robustly enough as his constant failures with a 775 board asserts.

 

[Abwx]

210W (average gaming minus idle) in ComputerBase's review at 'Max':

Nowhere near 135W.

We re talking of Furmark wich was used by two viral marketers in a video posted by a member here, and that s what is commented, so you re just bringing yet another irrelevant and goal post moving post...

They got in this bad shape through their own failures mostly. Consumers don't care about the financial situation of a company, they just want their product properly tested and engineered so it works properly.

It s much better enginered than the competition 700$ offering.

Compare the PSU and components and you ll see who is doing non proper enginering, given the prices the tables should had been turned, in the long run there s absolutely no doubt that the RX480 will be more reliable than both the GTXs 1080 and 1070.

 

[JDG1980]

210W (average gaming minus idle) in ComputerBase's review at 'Max':

The Maximized Power Target allows the power consumption of the Radeon RX 480 to rise significantly. 267 watts required for the graphics card in consistently applied 1,266 MHz, which additional 42 watts means.

Read that bit carefully. They managed to ram the power consumption up to 210W by increasing the power target. This is intentionally running the card out of spec, and not something the average novice user would do by accident.

The situation that concerns me most with this card is the possibility that it might damage someone's cheap OEM motherboard. That's the kind of situation where an ordinary user could plausibly run into trouble. I'd like to see a few real-world tests with crappy OEM boxes from Dell, HP, etc. and find out if that causes trouble.

IMO, running 3x video cards at full blast on non-powered risers constitutes abuse, and the fact that a motherboard slot burned out under that circumstance means nothing.

 

[Silverforce11]

Cool another data point. I didn't realize the ScienceStudio budget board was AM2. That's really old and it conforms to the idea that later versions of PCIE are built better than the 1.0 spec.

This guy was using an AM3 board that is hardly modern but still a more likely pick up for budget oriented users today. The reference can be handled better with today's motherboards even the cheaper ones. That said it's clear running over spec can cause issues if a board isn't designed robustly enough as his constant failures with a 775 board asserts.

Well if you're spec 2 and 3 for PCIE, you are supposed to handle up to 150/300W. Hence the extra ATX power connectors that were introduced at around this time as well as separate 12v rails to the CPU directly.

I'm surprised AMD didn't test on very old motherboards, it is an issue with two reviewers saying their RX 480 crash the system on AM2 and LGA 775 motherboards.

Though nothing blown, since they are designed to shut-down if they exceed limits. Except mining with non-powered risers on 3 cards, with a very old MB, that's suicidal.

 

[Abwx]

Read that bit carefully. They managed to ram the power consumption up to 210W by increasing the power target. This is intentionally running the card out of spec, and not something the average novice user would do by accident.

The situation that concerns me most with this card is the possibility that it might damage someone's cheap OEM motherboard. That's the kind of situation where an ordinary user could plausibly run into trouble. I'd like to see a few real-world tests with crappy OEM boxes from Dell, HP, etc. and find out if that causes trouble.

IMO, running 3x video cards at full blast on non-powered risers constitutes abuse, and the fact that a motherboard slot burned out under that circumstance means nothing.

The excess power is way too little to damage anything as if it fail at 80W then it will fail at 70W as well, there s no connectivity that could experience such step effects, it s like expecting that a 20A cable cant sustain 22A, that s just ridiculous.

Besides low quality MBs will provide less power through the PCIe connector by the virtue of their higher resistance in the MB supplies routings.

 

[3DVagabond]

The excess power is way too little to damage anything as if it fail at 80W then it will fail at 70W as well, there s no connectivity that could experience such step effects, it s like expecting that a 20A cable cant sustain 22A, that s just ridiculous.

Besides low quality MBs will provide less power through the PCIe connector by the virtue of their higher resistance in the MB supplies routings.

Old circuits to. We saw the pic of the guy with the burnt pci-e slot and there was obvious oxidizing on the slot.

 

[zinfamous]

I think a class action lawsuit is in order here. AMD engineering/QC teams are a joke. :thumbsdown:

Maybe once there is at least one piece of evidence where boards aren't being wrecked due to user error.

but for class action, you need way more than one case and there has yet to be even one case of a user frying their board with a 480X within normal use and spec.

so...gl with that.

 

[Abwx]

Old circuits to. We saw the pic of the guy with the burnt pci-e slot and there was obvious oxidizing on the slot.

We dont know from where it originate, what is sure is that he aknowledged using previously a GTX960 in this slot for FoldingHome and that he made some test with it at much higher power that what he usualy got.

What is sure is that a correctly designed slot shouldnt experience any oxydation over years as if it s of decent quality there s a few micrometers gold film that recover the copper contacts.

It s quite possible that some cheap connectors use silver wich conduct better but is subject to oxydation over the years, i cant tell at this point for PCs but it s used in high frequency circuitries to my knowledge.

 

[96Firebird]

One of those 8 pins is OPTIONAL, without it, it's 125W.... Not all PSUs utilize the 8th pin (which is a reserved Sense pin), and not all cards utilize it either.

Where did you learn this? All sources point to 75W slot power, 75W 6-pin power, and 150W 8-pin power being the PCI-E spec. Please back up your claim.

 

[sontin]

Read that bit carefully. They managed to ram the power consumption up to 210W by increasing the power target. This is intentionally running the card out of spec, and not something the average novice user would do by accident.

The average novice? AMD is promoting a new overclocking tool for the RX 480. :\

And why is the card running out of spec when you just increase the power limit?! Wouldnt it run within the specs because the "average novice" would expect that a +50% limit would put the card right at the 150W spot?!

And the last question: Why does the card and the software even allow to run outside the spec at all?