The 480: power consumption, PCI-E powerdraw

[antihelten]

This is disingenuous because only two aftermarket 960's were tested in this way.

No, seven aftermarket 960s were tested this way. Next time try and actually read the review.

Also, aftermarket cards are supposed to be safer, more over engineered, and better than reference cards. That an AIB 960 had this problem is more damning, not less. The AIB 480's are the ones people are looking to to correct the 480 problem!

Aftermarket cards aren't really supposed to be anything in particular, they can be whatever the AIB partner wants. There are plenty of bargain bottom aftermarket cards out there that are overall worse quality than the reference version. This tends to be particularly true for AMD, since they tend to over engineer their reference cards.

The point here is though, that the reference version will undoubtedly out sell any individual aftermarket version several times over, and as such it is a much bigger issue.

 

[bill3]

No, seven aftermarket 960s were tested this way. Next time try and actually read the review.

Editz: I was incorrect. However, there was still no internet hue and cry over just the Strix card.

Also, at least a couple more of those 960's seem to spike above 75 watts consistently, although nothing as bad as the Strix or the reference 480.

There were no reference 960's tested, so we dont know that picture.

Also not sure why Tom's separates out the 3.3V rail, where this Brazil site I looked at that tested the 480 was adding both the 12V and 3.3V together for PCI-E slot consumption. Adding the 3.3V rail would make all those 960's charts look worse.

I haven't looked into it greatly but a lot of people say the 750 Ti was also abusing the PCI-E slot well past 75 watts in the past regularly.

 

[24601]

All they need to do is to cap the PCI-e power to a maximum of 66w and stuff the rest of the power to the 6 pin. The 6 pin will still be able to handle the extra 20w of draw no problem.

You're not going to "software/firmware patch" hard-wired connectors and physics.

There is no distributor between the PCI-E slot power in and the PCI-E PSU connector (Just like most cards).

 

[antihelten]

There is no proof at all they tested all 7 960's AT THE PCI-E SLOT. In fact, they almost certainly didn't.

There is only PCI-E draw data presented for two cards. One failed. Period. Anything else is your speculation.

Again, read the freaking review, it's not that hard. Here I'll help you:

And AIB cards outsell reference cards by many times, unless OEM's buy all reference cards?

And who do you think ends up buying the OEM machines containing said cards?

 

[bill3]

My guess is of course, worse case they can downclock the card a bit in software to solve any issues. My question is how much to be safe? I dont mind losing a few percent of performance, especially considering the 480 will be like 10X faster than the card I'm coming from, and I dont overclock. But if it was say a 15% loss that'd be troubling. My guess is since power use tends to be logarithmic, even a small down clock will greatly ease any power issues.

My guess is AMD might issue such a driver/Bios update for those who absolutely want/need their card to be in spec, which then most people will ignore in favor of keeping higher clocks.

 

[Face2Face]

Here's an interesting video:

https://www.youtube.com/watch?v=rhjC_8ai7QA

Also, user issue over on Reddit. Who knows if true?

https://community.amd.com/thread/202410

 

[bill3]

We see the Gainward, Inno, and Palit is also spiking past 75. And spikes are more dangerous than stable draw.

At the least the Strix is in a huge way.

And if you need more examples I can start digging up 750 TI's.

 

[amenx]

Maybe 480 owners can downclock their cards until a solution shows up. Which brings me to a theory... I wonder if the 480 was meant to be released at 1080mhz stock clocks (as per initial leaks) which may not have had any power issues. And at a later stage, too close to manufacturing deadlines, bumped it to its current 1120mhz clocks without thorough testing.

 

[antihelten]

We see the Gainward, Inno, and Palit is also spiking past 75. And spikes are more dangerous than stable draw.

At the least the Strix is in a huge way.

And if you need more examples I can start digging up 750 TI's.

No spikes are not automatically more dangerous than stable overdraw. Whether or not spikes are more dangerous depends entirely upon the frequency and size of the spikes, and in the case of Gainward, Inno and Palit the spikes are just fine, unlike the ASUS card.

And by all means, show me some 750 Ti's that goes over the PCIe spec to the same degree that RX 480 or the ASUS 960 Strix does (PS. We already discussed the 750 Ti back on page 5).

 

[bill3]

No spikes are not automatically more dangerous than stable overdraw. Whether or not spikes are more dangerous depends entirely upon the frequency and size of the spikes, and in the case of Gainward, Inno and Palit the spikes are just fine, unlike the ASUS card.

You dont know that.

"And by all means, show me some 750 Ti's that goes over the PCIe spec to the same degree that RX 480 or the ASUS 960 Strix does (PS. We already discussed the 750 Ti back on page 5)."


Dont feel like it right now. Maybe later I'll try to dig that up. I've heard some draw up to 90 watts from the PCI-E slot.

Very few cards have been tested for PCI-e Slot draw is the issue. My suspicion is lots fail spec. Back in 750 Ti time I'm sure PCI-E slot data was even more rare if it existed at all.

 

[William Gaatjes]

Higher spikes is definitely worse for electronic equipment. If you don't believe me, go as an electrical engineer. My friend is one and spikes, along with tin splinters causes frequent damage to electronics is far worse than steady voltages, as long as it's not too far beyond rated capability.

Current spikes are not an issue if you stay within the limits of the component. Most semiconductor components like mosfets and diodes specify peak currents that can be for example 5 to 6 times higher than the continuous allowed current. As long as the spike is conform the allowed duty cycle.

Theoretically speaking :
On a motherboard however, current spikes through thin copper traces causes magnetic fields that can induce currents in nearby parallel traces that are connected to high impedance inputs and can cause weird effects. But that 12V line is certainly a big copper plane making it less of an inductor because of the geometry. But i can imagine that the 75W spec is certainly to prevent causing weird effects like crashes because some bit gets flipped because a voltage is induced in a nearby trace. But you would need to draw a lot more power than just 75W. At least double that amount i guess or even quadruple.

And tin whiskers are indeed a serious issue.

 

[antihelten]

You dont know that.

Don't know what? that having a few spikes of 20-30% above spec isn't an issue compared to constant spikes of 100-200%?

Dont feel like it right now. Maybe later I'll try to dig that up. I've heard some draw up to 90 watts from the PCI-E slot.

So you just offered to dig up some 750 Ti numbers, and then less than 15 minutes later you don't feel like it anymore, yeah right

And drawing 90 W is hardly comparable to the Strix that draws upwards of 250W or the RX 480 that draws up to 150W. Not to mention that both sit consistently above 75W.

 

[bill3]

90 W consistently...that would be higher than either...but I just read that in a comment somewhere, I dont have backing. And I dont feel like digging right now.

The more I look at this issue, I think I'm going to try to refuse shipment on my 480. Kills me but...I'm kinda OCD and it will just bother me...even though I know there will probably never be an issue.

It's too bad Nvidia is such a joke...3GB 1060? REALLY after all these years of shorting consumers on RAM? If they price the 6GB version at 249, I'll accept nothing less...it's that or some other crazy stupid waiting weeks and weeks for 470 or AIB 480's.

Thing is I have a 2 week vacation coming July 12 and I really want a new card to play with by then

I wish I could see the future, if I knew the 6GB 1060 would be priced at 249, I'd wait for that. If it's 299 that's too much.

 

[Doom2pro]

And tin whiskers are indeed a serious issue.

I thought tin whiskers were limited to pure tin solder? Isn't the point of the solder additives to prevent tin related issues (especially the expansion of tin under sub zero temperatures).

 

[antihelten]

90 W consistently...that would be higher than either...but I just read that in a comment somewhere, I dont have backing. And I dont feel like digging right now.

Well, let me know when you actually have time to back up your arguments with data then. I'll wait.

 

[bill3]

You mean like I already have multiple times such as the Strix 960?

 

[Stuka87]

Maybe 480 owners can downclock their cards until a solution shows up. Which brings me to a theory... I wonder if the 480 was meant to be released at 1080mhz stock clocks (as per initial leaks) which may not have had any power issues. And at a later stage, too close to manufacturing deadlines, bumped it to its current 1120mhz clocks without thorough testing.

No need to down clock, you just set the power limit to something like -10%, and that will effectively drop the power usage. Or, do as some others have, and undervolt, that drops the power usage a lot.

 

[Azix]

the question is whether the constacts on the pci slot can handle it.

they are rated for 1.1A in the pci-e 1.1 specs. there are 5 of these 12V contacts leading to the 5.5 A.

But they are 1.1 A minimum and the excess even with OC is something like 0.4 A per pin. They are also supposed to not reach 30 degrees above ambient at that current, but are also heat tested up to 105 degrees for several hours. seems they should be fine, but if there is resistance on some of these contacts (dust etc.) maybe there could be damage.

 

[sirmo]

Current spikes are not an issue if you stay within the limits of the component. Most semiconductor components like mosfets and diodes specify peak currents that can be for example 5 to 6 times higher than the continuous allowed current. As long as the spike is conform the allowed duty cycle.

Theoretically speaking :
On a motherboard however, current spikes through thin copper traces causes magnetic fields that can induce currents in nearby parallel traces that are connected to high impedance inputs and can cause weird effects. But that 12V line is certainly a big copper plane making it less of an inductor because of the geometry. But i can imagine that the 75W spec is certainly to prevent causing weird effects like crashes because some bit gets flipped because a voltage is induced in a nearby trace. But you would need to draw a lot more power than just 75W. At least double that amount i guess or even quadruple.

And tin whiskers are indeed a serious issue.

Big spikes can cause loop stability issues as well as propagate transients into other components, a slow acting 3.3V VRM can become unstable and overvolt the components on the 3.3 rail.. transients are a bigger problem imo. A spike can also cause overvoltage issues from the PSU which is too slow to react and is trying to compensate, damaging the filtering caps for instance.

 

[Doom2pro]

Big spikes can cause loop stability issues as well as propagate transients into other components, a slow acting 3.3V VRM can become unstable and overvolt the components on the 3.3 rail.. transients are a bigger problem imo. A spike can also cause overvoltage issues from the PSU which is too slow to react and is trying to compensate, damaging the filtering caps for instance.

This is why good designs have beefy bypass capacitors in such areas... If voltage ripple was that devastating, people using cheap power supplies would be complaining far more often of dead motherboards and or other components.

Funny thing is, these cheap power supplies can work at a minimum for quite a while, with problems being system instability.

Accelerating aging of semiconductors on the other hand is a real concern with high voltage ripple, but I don't think these VRMs are going to be that sensitive to ripple, their max specifications are derated for this reason.

 

[bononos]

You mean like I already have multiple times such as the Strix 960?

The Strix960 is the only problematic model but at least it the average draw doesn't exceed pcie specs. Having a RX480 reference model fail is clearly alot worse.

The 750ti power issue you mentioned is over here:
http://www.tomshardware.com/reviews/geforce-gtx-750-ti-review,3750-20.html

 

[bill3]

For the millionth time, there was no average power draw from the PCI-E slot for the Strix established...it is simply not mentioned anywhere in Tom's testing.

Anyways I did the deed, was able to refuse shipment on my 480 (it had already been delivered, but you can take it to any fed ex office after the fact and refuse it as long as it's not opened). Newegg already told me this would result in a full refund (minus my $17 of shipping fees I assume).

Kinda sucks returning a powerful, cheap, and popular card that is sold out everywhere, but basically this lets me keep my powder dry and options open for a couple weeks. Either if the 6GB 960 is priced at 249 or less, or if a suitable fix for the 480 is announced in the next couple weeks, I can jump on either option. $177 down the drain sucks, but I would have probably lost more like $100 had I opened the 480 and later ended up ebaying it.

 

[bill3]

The Strix960 is the only problematic model but at least it the average draw doesn't exceed pcie specs. Having a RX480 reference model fail is clearly alot worse.
QUOTE]

An AIB model failing is worse IMO. The AIB models are the very ones people are looking to to FIX the 480 problem. AIB always have better quality and components than reference, that's why they exist, whereas reference is the lowest priced version. Most people expect AIB 480 to ship with minimum 8 pin connector for example, where the reference ships with a weaker 6 pin connector.

 

[MrTeal]

The current spikes aren't an issue for the connector. Voltage spikes could be as high voltage can cause arcing, but we're well away from the kinds of voltages where you'd worry about that. Current spikes can cause issues with electronics, but this isn't one of those times. Here you're looking at heat causing an issue, and that's all determined by average power.

 

[StrangerGuy]

Somebody here really haves a bad case of false equivalences.

One single AIB 960 spiking above 75W but average intended for aftermarket retail = RX480 ref that draws >75W average and even worse spiking on a huge % of cards designed not only for retail but also white box OEMs.

So easy to claim everything is A-OK when one doesn't have to bear any responsibility. I maybe will take him seriously only if he agrees to pay for any board fried by the ref RX480 aka walk the bloody talk.