Discussion: Polaris, AMD's 4th Gen GCN Architecture

[Rogue Leader]

Ah, for some reason I confused that with Sapphire. It can't just pop out, though, without unplugging the cable for it.

Look at the pic of the fan removed, on the bottom right, the connector is integrated into the bottom clip.

 

[turkey3_scratch]

Ah, for some reason I confused that with Sapphire. It can't just pop out, though, without unplugging the cable for it.

Look at the pic of the fan removed, on the bottom right, the connector is integrated into the bottom clip.

Well, it's nice and all, but I wonder how much a replacement fan would cost, and I wonder how reliable these fans are, what bearing they have.

 

[Rogue Leader]

Ah, for some reason I confused that with Sapphire. It can't just pop out, though, without unplugging the cable for it.

Look at the pic of the fan removed, on the bottom right, the connector is integrated into the bottom clip.

Well, it's nice and all, but I wonder how much a replacement fan would cost, and I wonder how reliable these fans are, what bearing they have.

Probably more than a normal fan of that size, $30 maybe?

TBH Its rare that they need to be replaced, but this will make cleaning easier, thats about it.

 

[littleleo]

The RX480 doesn't burn motherboards. Wccftech deleted the article.

Then why is AMD coming out with a fix if there is nothing to fix?

Nvidia's astroturfers said that Crimson drivers burned GPUs. False.

I'm sorry I missed that when did this issue pop up?

Nvidia's astroturfers said that the RX480 burns mobos. False.

I believe it was people that rushed out and bought the RX-480 that reported the issue not Nvidia "astrturfers" most likely AMD enthusiasts.

The 960 has higher peaks than the 480.

Perhaps funny how no one ever complained about it burning up a motherboard.

Nvidia's cheating benchmark since 2003.

I think they all cheat a little that is why it is always good to get multiple bench marks from different sites before you make a decision.

I wouldn't be surprised if the issue isn't has bad as it was 1st reported. In reading the post of those that reported the issue they seem few in number and seem to have been OCing the card. If I had one of these cards I think I'd just not OC it for now until the new driver comes out. For the money I still think this is a heck of a card. AMD has always had several good options for the hole between the GTX 960 and the GTX 970. Perhaps Nvidia's new GTX 1060 will finally address that segment to combat the RX-480.

 

[littleleo]

I like XFX's designs, too. Can't wait to see what their aftermarket card looks like.

We have seen it, it even has replaceable fans. Pic was included in instruction booklet of the reference cards

That does look cool. XFX has always had some good stuff.

Do you think XFX regrets dumping Nvidia for AMD? Going to a chip brand that has seen it's market share drop to 20%. I wonder if the guy that recommended that and the person that followed his advice is still working there.

 

[TJ Hooker]

The RX480 doesn't burn motherboards. Wccftech deleted the article.

Then why is AMD coming out with a fix if there is nothing to fix?

Acknowledging that the 480 draws more power than is allowed by the PCIe spec (and fixing this) is not the same as acknowledging that the 480 burns motherboards.

 

[jimmysmitty]

The RX480 doesn't burn motherboards. Wccftech deleted the article.

Then why is AMD coming out with a fix if there is nothing to fix?

Acknowledging that the 480 draws more power than is allowed by the PCIe spec (and fixing this) is not the same as acknowledging that the 480 burns motherboards.

Correct.

However there is a correlation between the two as mentioned in the follow up article, the extra watts are not what is concerning but rather the current, which they noted was 23% above spec and high current can cause potential damage.

All this says is that it is a possibility that the RX 480 pulling more power than needed from the PCIe slot is possibly causing it to burn the PCIe slots or damage the motherboard.

Think of it this way, the FX 9590 is not burning up AM3+ motherboards but rather it draws more power than the boards that do have issues with it allow that can cause damage to the board.

 

[Mousemonkey]

Nvidia's astroturfers said that Crimson drivers burned GPUs. False.

AMD released Radeon Software Crimson 15.11.1 Update. This update fixes an important issue with VGA fan speed control that overheated GPUs in certain RPM conditions as well as adding fixes for Just Cause 3, Fallout 4, Star Wars: Battlefront, and Call of Duty Black Ops III.

http://www.guru3d.com/news-story/amd-releases-radeon-software-crimson-edition-15-11-1-hotfix-update.html

I really don't know where you are getting your information from mate, but it appears to be incorrect.

 

[Sakkura]

So you are saying it is all just a coincidence?

It is still higher than spec power draw and AMD could have avoided all of this by just adding a 8pin connection to it instead so the PCIe port would draw less power.

Why are you refusing to read my posts? I specifically said that the issue with the card was exposing an issue with the motherboard.

A motherboard issue? No it's an issue with the card. You would expect that any consumer-based product you purchase to stick into a motherboard would comply with the specs and standards.

As I explained in my post, and as the Tom's Hardware article also indicates, an RX 480 in itself will not cause fried motherboards. So when you have an example of a fried motherboard, then part of the blame is also on the motherboard. There was a flaw in the motherboard, and the flaw in the graphics card exposed that flaw.

 

[Sakkura]

So you are saying it is all just a coincidence?

It is still higher than spec power draw and AMD could have avoided all of this by just adding a 8pin connection to it instead so the PCIe port would draw less power.

Why are you refusing to read my posts? I specifically said that the issue with the card was exposing an issue with the motherboard.

I am not. I just don't think it is the motherboard that is the issue I think it is the GPUs power draw. If this happened with nVidia I would be on the same page it is just odd that people who ran older, higher power draw cards did not have this issue because those GPUs did not pull as much power from the PCIe slot.

It is too much of a coincidence.

Oh FFS. Just read one of my replies, I've explained the concept perfectly well. There isn't necessarily a single cause of such a problem.

If you run current through a wire, and the wire burns, whose fault is it? The wire for not being thick enough, or the current for being too strong? It's the combination.

What we have in this situation is a graphics card pulling more current than the specification allows. But we also know that the current the graphics card pulls will not cause a properly constructed motherboard to sustain damage. The articles have explicitly said so. So the graphics card issue exposed a pre-existing issue in the motherboard. This is not a coincidence, as you keep erroneously claiming, it is in fact cause and effect.

 

[BurgerandChips66]

I would expect that as soon as possible, AMD will ensure that all reference cards they manufacture do not draw excessive power via the PCIe slot or draw over ~150W total (at stock boost) regardless of whether or not the latest driver (with the fix) is present.

 

[Math Geek]

not sure why you folks are letting these random people talk you in circles over and over. as soon as one of the trolls leaves another comes along and you repeat the cycle.

just give it up. the fanboys are trolling and they caught you hook, line and sinker. you're trying to use facts and logic with a totally illogical person who is deliberately avoiding the facts. i'd love to see this thread make it back to some intelligent conversation about the cards again, rather than a feeding place for every nvidia troll on the forum!!!

 

[TehPenguin]

The reference PCB design of the RX 480 includes a surprisingly over-engineered voltage regulation and power delivery circuitry. The reference PCB features 6+1 power phases capable of delivering more than 58% higher current compared to the 5+1 design of the Nvidia GTX 1080 Founder’s Edition. Enabled primarily by higher quality inductors and an additional power phase. Each phase also features twice the number of mosfets per inductor compared to the GTX 1080.

Can someone explain this part? How can a higher current benefit Polaris, more wattage? Could this be the reason for the higher than TDP power draw?

I was THIS close to pulling the trigger on a reference 480. I just need a 1080p card for a year or so. Now that aftermarket 480's are supposedly coming mid July I find myself waiting, AGAIN.

If this goes on it'll be 2020 before I upgrade.

Send help D:

 

[Suzuki@1473780084@1473780163]

Sakkura don't bother...

 

[cdrkf]

not sure why you folks are letting these random people talk you in circles over and over. as soon as one of the trolls leaves another comes along and you repeat the cycle.

just give it up. the fanboys are trolling and they caught you hook, line and sinker. you're trying to use facts and logic with a totally illogical person who is deliberately avoiding the facts. i'd love to see this thread make it back to some intelligent conversation about the cards again, rather than a feeding place for every nvidia troll on the forum!!!

It's quite sad how many people just want to see AMD fail and die- I guess it's the same mentality that drives many football fans in the UK to support whichever team they believe is strongest at the time, regardless of having any affiliation with said team (like the thousands of 'Manchester United' supporters who have never been to Manchester in their lives).

At the moment, AMD is in a weak position- so out they come. The 1060 has already been declared to be far superior and 'the winner' before we've seen benchmarks, pricing and so on. Whilst i agree 'don't feed the trolls' is a good philosophy you do have to think- if they are allowed to control the conversation then THEIR version of reality is the one that will take hold. Every graphics card discussion on the forums will be 'AMD is rubbish, buy nVidia'... which is invariably going to hurt AMD sales more. I think it's important to keep some balance. Graphics cards are an internal PC component- yet the manufacturers are trying to make them into a kind of fashion item and form a following. This is very bad for the consumer in the long run as its trying to stop people buying based on logic. Where a component is concerned, people should be buying based on requirements, not due to brand or that it has fancier packaging (or RGB lighting, seriously?).

 

[TJ Hooker]

The reference PCB design of the RX 480 includes a surprisingly over-engineered voltage regulation and power delivery circuitry. The reference PCB features 6+1 power phases capable of delivering more than 58% higher current compared to the 5+1 design of the Nvidia GTX 1080 Founder’s Edition. Enabled primarily by higher quality inductors and an additional power phase. Each phase also features twice the number of mosfets per inductor compared to the GTX 1080.

Can someone explain this part? How can a higher current benefit Polaris, more wattage? Could this be the reason for the higher than TDP power draw?

I'm guessing this relates to the following:

We’ve read here and there that AMD is using a much larger MOSFET than would be necessary on the low side (GPU side) with the MDU1511. The argument is that this results in massive overclocking potential.
This component is capable of handling up to 100A, which means that every phase could be pushed to extreme limits, beyond 100W.

From:http://www.tomshardware.com/reviews/amd-radeon-rx-480-power-measurements,4622-2.html
Higher current capacity means the VRMs can supply more power, which is required for overclocking. But considering the limited power connectors and power distribution issues with the reference design, it may not be that beneficial for overclocking. It shouldn't increase power draw at stock clocks though. May even reduce it:

The reason that AMD chose this particular MOSFET is probably completely different, though: it has a very low internal resistance of just 2.4 mΩ. This minimizes loss in the power converters, resulting in both a cooler card and a lower power consumption.

 

[TehPenguin]

Spoiler

The reference PCB design of the RX 480 includes a surprisingly over-engineered voltage regulation and power delivery circuitry. The reference PCB features 6+1 power phases capable of delivering more than 58% higher current compared to the 5+1 design of the Nvidia GTX 1080 Founder’s Edition. Enabled primarily by higher quality inductors and an additional power phase. Each phase also features twice the number of mosfets per inductor compared to the GTX 1080.

Can someone explain this part? How can a higher current benefit Polaris, more wattage? Could this be the reason for the higher than TDP power draw?

I'm guessing this relates to the following:

We’ve read here and there that AMD is using a much larger MOSFET than would be necessary on the low side (GPU side) with the MDU1511. The argument is that this results in massive overclocking potential.
This component is capable of handling up to 100A, which means that every phase could be pushed to extreme limits, beyond 100W.

From:http://www.tomshardware.com/reviews/amd-radeon-rx-480-power-measurements,4622-2.html
Higher current capacity means the VRMs can supply more power, which is required for overclocking. But considering the limited power connectors and power distribution issues with the reference design, it may not be that beneficial for overclocking. It shouldn't increase power draw at stock clocks though. May even reduce it:

The reason that AMD chose this particular MOSFET is probably completely different, though: it has a very low internal resistance of just 2.4 mΩ. This minimizes loss in the power converters, resulting in both a cooler card and a lower power consumption.

+1, thanks.

 

[BurgerandChips66]

https://youtu.be/plC7tOYIqBw?t=3240

If you can be bothered to watch this rather long geeky video (!) particularly from about 54 minutes onwards, it explains what went wrong with the RX 480 reference card design - in terms of why there are issues with the power draw and the distribution of that power draw.

I hope and expect that this is something the custom cards will solve.

Spoiler: It's inherant to AMD's configuration of the PCB power distribution and particularly their configuration of the 6-pin socket.

 

[Suzuki@1473780084@1473780163]

How is it inherent when AMD officially said that it's not?

 

[Ags1]

For me one good thing about "powergate" was that it caused me to cancel my order for a reference RX480. I wanted to hold out for the custom cards but my willpower failed - there I was clicking Buy on launch day. But if I can pick up a card clocked at 1400+ MHz and under a TDP of 200W I would be really happy!

 

[BurgerandChips66]

From what I could gather in the video, it's inherant to AMD's own choice of configuration of the PCB power distribution and their own choice of configuration for the 6-pin socket. Rather than simply powering the GPU core soley from the 6-pin and the vram from the PCIe slot everything was shared.

Maybe someone who understands the standards and the accepted practice can watch the video and explain it better!

These videos are also interesting and much shorter https://youtu.be/qG2e-v94L4M https://youtu.be/Jq47qmwcus8

 

[jimmysmitty]

So you are saying it is all just a coincidence?

It is still higher than spec power draw and AMD could have avoided all of this by just adding a 8pin connection to it instead so the PCIe port would draw less power.

Why are you refusing to read my posts? I specifically said that the issue with the card was exposing an issue with the motherboard.

I am not. I just don't think it is the motherboard that is the issue I think it is the GPUs power draw. If this happened with nVidia I would be on the same page it is just odd that people who ran older, higher power draw cards did not have this issue because those GPUs did not pull as much power from the PCIe slot.

It is too much of a coincidence.

Oh FFS. Just read one of my replies, I've explained the concept perfectly well. There isn't necessarily a single cause of such a problem.

If you run current through a wire, and the wire burns, whose fault is it? The wire for not being thick enough, or the current for being too strong? It's the combination.

What we have in this situation is a graphics card pulling more current than the specification allows. But we also know that the current the graphics card pulls will not cause a properly constructed motherboard to sustain damage. The articles have explicitly said so. So the graphics card issue exposed a pre-existing issue in the motherboard. This is not a coincidence, as you keep erroneously claiming, it is in fact cause and effect.

And if a company forces a higher current than the wire is rated or speced for whose fault is it? The motherboards are designed within the PCIe spec. The GPU is pushing it beyond that spec.

In short you are saying that any motherboard (so far ASRock, Asus and an MSI) are all faulty and at fault for a GPU pulling more power consistently than the slot is speced for?

I just can't see how a motherboard can be called faulty if a part is going beyond spec.

 

[TehPenguin]

https://youtu.be/plC7tOYIqBw?t=3240

If you can be bothered to watch this rather long geeky video (!) particularly from about 54 minutes onwards, it explains what went wrong with the RX 480 reference card design - in terms of why there are issues with the power draw and the distribution of that power draw.

I hope and expect that this is something the custom cards will solve.

Spoiler: It's inherant to AMD's configuration of the PCB power distribution and particularly their configuration of the 6-pin socket.

Thanks! That was actually really interesting. From what I understood AMD does not utilize the 6pin fully. Might this be because they don't need to?(as their power phases are capable of delivering more current)

Will have to dive into the topic a bit more.

 

[17seconds]

AMD was trying to shed the image that their cards are power hogs. Using an 8 pin plug in the age of next gen 14nm FinFET would have kept that perception alive, particularly when sitting next to Pascal.

 

[Suzuki@1473780084@1473780163]

So you are saying it is all just a coincidence?

It is still higher than spec power draw and AMD could have avoided all of this by just adding a 8pin connection to it instead so the PCIe port would draw less power.

Why are you refusing to read my posts? I specifically said that the issue with the card was exposing an issue with the motherboard.

I am not. I just don't think it is the motherboard that is the issue I think it is the GPUs power draw. If this happened with nVidia I would be on the same page it is just odd that people who ran older, higher power draw cards did not have this issue because those GPUs did not pull as much power from the PCIe slot.

It is too much of a coincidence.

Oh FFS. Just read one of my replies, I've explained the concept perfectly well. There isn't necessarily a single cause of such a problem.

If you run current through a wire, and the wire burns, whose fault is it? The wire for not being thick enough, or the current for being too strong? It's the combination.

What we have in this situation is a graphics card pulling more current than the specification allows. But we also know that the current the graphics card pulls will not cause a properly constructed motherboard to sustain damage. The articles have explicitly said so. So the graphics card issue exposed a pre-existing issue in the motherboard. This is not a coincidence, as you keep erroneously claiming, it is in fact cause and effect.

And if a company forces a higher currant than the wire is rated or speced for whose fault is it? The motherboard are designed within the PCIe spec. The GPU is pushing it beyond that spec.

In short you are saying that any motherboard (so far ASRock, Asus and an MSI) are all faulty and at fault for a GPU pulling more power consistently than the slot is speced for?

I just can't see how a motherboard can be called faulty if a part is going beyond spec.

No, AMD proved that the motherboards are too weak, not that they are outside spec, that's already known. Tom's already explained that the RX480 shouldn't be able to damage the motherboards, did you even read the article?