AMD releases a new statement regarding the RX 480's power issues

[WYP]

AMD releases a new statement regarding the RX 480's power issues, stating that they will be releasing a new driver within the next 48 hours which will lower PCIe power draw and enhance performance by up to 3%.

Read more on AMD's upcoming 16.7.2 driver.

 

[MathiasLindberg]

AMD said they were going to keep improving on the RX 480, and they did. They also fixed the issue with power distribution AND provided the option to draw a lot less power for minimal negative, or even a positive performance change.

 

[NeverBackDown]

They are more than likely to switch where the power comes from. Instead of the slot, transfer it to the 6pin.. Which is fine as nearly all are overbuilt for extra power draw anyway and if it gets to dangerous, it has over current protection, unlike the motherboard. But I doubt it. Just look at the 295x2. That drew more than what it was rated for, significantly more too

 

[RobM]

I'm more concerned about them wiring the 6 pin like it was 8 pin and how its totally isolated from the pci express side the middle pin should be used to detect plug yet it is actually a ground and thus capable of dealing with over 200w

 

[Thelosouvlakia]

Regarding this whole thing people called "Powergate" first of all, at least it was an oversight on AMD's side and actually fixable (at least what they say for now) with a driver/bios update. Good for them. At least we can suppose it wasn't on their knowing that this issue would occur, like NVIDIA did with the 970.

How do we know that drawing that much power through the PCIe was such a bad thing, most motherboards out there are designed to handle SLI and crossfire and should have circuitry able to handle the powerdraw without getting damaged.

Anyway..

 

[RobM]

I would not call the way the used the 6 pin as an oversight at all. it is not connected to the correct standard.

 

[RobM]

the 6 pin on the rx480 is only powering the vcore vrm its connected out of spec

 

[jmaster299]

Regarding this whole thing people called "Powergate" first of all, at least it was an oversight on AMD's side and actually fixable (at least what they say for now) with a driver/bios update. Good for them. At least we can suppose it wasn't on their knowing that this issue would occur, like NVIDIA did with the 970.

How do we know that drawing that much power through the PCIe was such a bad thing, most motherboards out there are designed to handle SLI and crossfire and should have circuitry able to handle the powerdraw without getting damaged.

Anyway..

Not true about the PCIe bus being designed to handle extra power. Each PCIe x16 slot has 5 pins that deliver power from the 12v rail. Each pin is rated to provide only 1.1 amps each. That is where the 66w power limit comes from, watts are power that is made up from amps.

The 66w on the 12v rail that the PCIe bus is rated to handle comes each of those 1.1a pins supplying 13.2w each. When the PCIe bus was supplying a constant draw of over 66w in testing that means it was supplying more than a total of 5.5a, which means it was supplying more than 1.1a per pin.

Exceeding that amperage will absolutely those pins eventually. Go watch this video, the entire thing, and learn something. Stop listening to AMDs self serving lies.

https://youtu.be/ZjAlrGzHAkI

 

[NeverBackDown]

It is up to standard. If it wasn't they would not have been able to release it. It's just when any sort of OC is done (boosting including) it goes over spec. But it's really not a big deal. They are fixing it with two options, one is safer than current option and the other totally fixes it. Powergate is over... Partner cards won't even have this issue since most will have more power available

 

[RobM]

it is not to spec for a 6 pin and using bios to lower things thus performance is not really a fix

 

[NeverBackDown]

it is not to spec for a 6 pin and using bios to lower things thus performance is not really a fix

Should reread my posts.
As I have said, PSUs can handle more load on a 6pin, it's not a big deal, especially with overcurrent protection. That's one option they are allowing, to preserve motherboards that can not handle this, and no motherboard has overcurrent protection. The other fix they have totally negates any "over spec" for motherboard or 6pin by lower available power. So yes it is a fix. Performance more than likely won't change. Seeing as underclocking doesn't change anything, lowering volts and keeping the same clocks will yield the same results but at better power consumption levels that are in spec. If they didn't release it and applied this update, you wouldn't even notice. It's just extra tuning on there part that they overlooked/didn't anticipate.

 

[SuB]

It is up to standard. If it wasn't they would not have been able to release it. It's just when any sort of OC is done (boosting including) it goes over spec. But it's really not a big deal. They are fixing it with two options, one is safer than current option and the other totally fixes it. Powergate is over... Partner cards won't even have this issue since most will have more power available

Should reread my posts.
As I have said, PSUs can handle more load on a 6pin, it's not a big deal, especially with overcurrent protection. That's one option they are allowing, to preserve motherboards that can not handle this, and no motherboard has overcurrent protection. The other fix they have totally negates any "over spec" for motherboard or 6pin by lower available power. So yes it is a fix. Performance more than likely won't change. Seeing as underclocking doesn't change anything, lowering volts and keeping the same clocks will yield the same results but at better power consumption levels that are in spec. If they didn't release it and applied this update, you wouldn't even notice. It's just extra tuning on there part that they overlooked/didn't anticipate.

I'd bet a pretty penny your response to this would be polar opposite if this had been nVidia doing it...

Spec is spec for a reason. They release the card with a built in boost feature that takes it out of spec. That means, it's out of spec.

This kind of crap should have (probably was) shown up in testing of pretty much any kind. Honestly if anyone thinks that they didn't know about this... they're being a bit naive.

 

[RobM]

Should reread my posts.
As I have said, PSUs can handle more load on a 6pin, it's not a big deal, especially with overcurrent protection. That's one option they are allowing, to preserve motherboards that can not handle this, and no motherboard has overcurrent protection. The other fix they have totally negates any "over spec" for motherboard or 6pin by lower available power. So yes it is a fix. Performance more than likely won't change. Seeing as underclocking doesn't change anything, lowering volts and keeping the same clocks will yield the same results but at better power consumption levels that are in spec. If they didn't release it and applied this update, you wouldn't even notice. It's just extra tuning on there part that they overlooked/didn't anticipate.

Should read my posts fella
They ARE NOT TO SPEC
"overcurrent protection" in essence there is no protection in the way AMD have implemented the 6 pin on the RX card so no it is not a fix it is a workaround that is easier than recalling a product.
With these cards it is in theory possible to blow the output circuit most likely the capacitors would be the things that go.

I really like AMD stuff but to implement a card in this way is just plain dumb

so not only do we have a card thats out of spec in regards to the 6 pin (and potentially Fatal to the card) it is also drawing out of spec for the 12v pci lane with a "software fix" that has functionality turned off

 

[Kushiro]

They got away with it for the 295x2, maybe they thought they would try their luck.

Seriously though, with a quality board and psu this aint a problem. But hey, thats not everyone. Especially not the budget oriented crowd this card is for. Many would have decent hardware.

 

[NeverBackDown]

I'd bet a pretty penny your response to this would be polar opposite if this had been nVidia doing it...

Spec is spec for a reason. They release the card with a built in boost feature that takes it out of spec. That means, it's out of spec.

This kind of crap should have (probably was) shown up in testing of pretty much any kind. Honestly if anyone thinks that they didn't know about this... they're being a bit naive.

Bet as many pennies as you want, you'd still be wrong. It's in spec. Whether it's AMD or Nvidia is doesn't change anything. However once it boosts it becomes out of spec, all they would really need to do is remove the boost or lower it. Instead they are opting for other changes, whether it's good or bad remains to be seen. It goes through multiple standard tests, if it passed all of them, then it won't be a big problem and there is leeway in these standards and it more than likely fell into that area. But again, long term we don't know and probably won't.

Now what I will say against Nvidia is at least AMD didn't try to cover it up for month's or even tell reviewers about the 970. Did it make a difference to performance? No. And I doubt this does either with the driver fix