AMD Releases Statement On Radeon RX 480 Power Consumption; More Details Tuesday
by Ryan Smith on July 2, 2016 5:05 AM EST
Following up on this week's Radeon RX 480 launch, there has been some questions raised about the power consumption of the card. This is after some sites whom directly tap the power rails feeding the card discovered that at least some of their samples were pulling more than the standard-allowed 75W over the PCIe slot and/or 6-pin PCIe external power connector.
To that end, it would appear that AMD's staff is working weekend duty, and they have just sent over the following statement.
As you know, we continuously tune our GPUs in order to maximize their performance within their given power envelopes and the speed of the memory interface, which in this case is an unprecedented 8Gbps for GDDR5. Recently, we identified select scenarios where the tuning of some RX 480 boards was not optimal. Fortunately, we can adjust the GPU's tuning via software in order to resolve this issue. We are already testing a driver that implements a fix, and we will provide an update to the community on our progress on Tuesday (July 5, 2016).
If some of the data is to be believed, these cards are exceeding 150W total at times, which would mean there is either something causing them to run in the wrong power state, or they are just outright exeeding their power limit and need to be throttled back. As we don't do per-rail testing I don't have anything meaningful to add at this second, but it will be very interesting to see how AMD responds next week.
Update 07/06: AMD has since released their status update, which you can find here.
Facebook
Twitter
RSS
181 Comments
View All Comments
FriendlyUser - Saturday, July 2, 2016 - link
Slight undervolt, reduce power limit of the card so it hits max boost slightly less often. Strict compliance will be assured but at the cost of maybe 2-5% less performance.I highly suspect the 8GB of fast GDDR5 is the culprit. They should have stayed with 4GB only. No need for 8GB with the resolutions this card can manage.
jameskatt - Saturday, July 2, 2016 - link
Since the AMD GPU doesn't have an external power supply like nVidia uses, will it fry your motherboard and potentially your expensive Intel CPU?Oxford Guy - Saturday, July 2, 2016 - link
I'm sure it will -- just before the sky falls.D. Lister - Saturday, July 2, 2016 - link
It COULD fry a cheap motherboard, or even a decent one if overclocked. The CPU getting damaged is unlikely. Although I wouldn't want to put an SSD in such a system without making sure the SSD has power failure protection. Still, AMD says it will be fixed in a couple of days, so let's see how it turns out.HollyDOL - Sunday, July 3, 2016 - link
Tbh AMD made terrible decision with 6pin power plug... If they just plugged 8pin and said 150W average with peaks under stress at 170W, all would be fine, at least they wouldn't be overloading other components.D. Lister - Sunday, July 3, 2016 - link
An 8-pin connector would certainly be welcome, especially for a stable OC, but all by itself it won't be able to solve the 480's problem.The core issue here is not a new one for AMD - bad software, in this case, at BIOS level. They (AMD) are hinting that a simple driver patch would fix this, but they may eventually have to offer a new BIOS version altogether, with proper voltage distribution and more pragmatic speeds.
At the end of the day, the 480 could actually end up exceeding Maxwell in performance/watt (irony), while having to settle somewhere between the 960 and the 970 in raw performance. The questions that are bothering me at this point are: how much of the 480's MSRP would have to be sacrificed over this? And what impact that sacrifice would have on RTG's future plans?
HollyDOL - Sunday, July 3, 2016 - link
Let's think about it...Unless there is a BIOS patch that would rewrite power profiles drivers themselves won't be able to do that much - likely scenario is lowering max frequencies, but I am not confident it would be able to take out 15W of the budget. So maybe we are looking on combination of reduced voltages and frequencies a little. This is probably the least hurting scenario.
The question remains though... why did they take that much over PCIe in the first place... yep, 6pin is 75W capped, but practically majority of PSUs today have single rail for whole 12V, unless you had a very bad one, 5W more per such a big pin would be hardly a stress, even when it is above specs. Otoh, I am sure I wouldn't want that overload running over my motherboard, it's probably in safety margin, but we are talking about long term continuous stress which is asking for troubles. And honestly, what average gamer does when he gets all shiny new toy to play with? Put it in, take vacation and game a week in row, long term stress provided :-) I still wonder how they passed specification tests certification for the card. What worries me is the chance the excessive power load ran via PCIe was driven through there intentionally since it's much harder to detect there than simple 6/8pin power cord, AMD knowing about the issue and sneaking it in with "we'll fix with next driver and they'll never notice"
D. Lister - Sunday, July 3, 2016 - link
"The question remains though... why did they take that much over PCIe in the first place... yep, 6pin is 75W capped, but practically majority of PSUs today have single rail for whole 12V, unless you had a very bad one, 5W more per such a big pin would be hardly a stress, even when it is above specs."Of course, the 6p/8p PEG cables take the power straight from the PSU and their (comparatively) much thicker twisted copper wires have a fairly massive headroom of hundreds of watts. The PCI express slots are primarily a data transfer interface, and take their power from the very fine traces printed on the motherboard. These parts are not made for sustaining such electrical abuse, and every hardware manufacturer is expected to know this very basic stuff.
"I still wonder how they passed specification tests certification for the card."
There are always loopholes to be found to circumvent standards and regulations, if one is motivated enough to look for them. Hopefully there would be necessary amendments and added vigilance, to ensure nobody tries something like this again.
"What worries me is the chance the excessive power load ran via PCIe was driven through there intentionally"
The only rational alternative is that engineers at AMD are unaware of the basics of electronics, which is far less likely. Probably the seriousness of consequences of this design choice didn't travel up the chain of command with the necessary impact.
"AMD knowing about the issue and sneaking it in with "we'll fix with next driver and they'll never notice""
...or, "we'll gradually normalize it in the later production batches, and by the time there is a noticeable difference in framerates, Vega would be ready for launch and the voices of the naysayers would go unheard in the festivities." Besides, even if someone pointed to a performance decrease and didn't get immediately destroyed by the fanboys, AMD could always find a scapegoat, like some Windows update, or developers favoring Nvidia, or this or that or the freakin' aliens. :(
blzd - Saturday, July 2, 2016 - link
Any type of "fix" for this would need to reduce performance right? To lower the power draw somewhat.Is it safe to say that AMD is somewhat cheating their power numbers to make them look more efficient?
I believe it's safe to say nVidia is doing the same when their manufacturers list their cards (GTX 970 and 980) by only their reference power usage when they are in fact highly overclocked cards using multiple power connects and drawing much, much more.
Oxford Guy - Saturday, July 2, 2016 - link
My assumption is that fixing a problem with transient spikes is hardly in the same league as the 970 VRAMgate.The market rewarded Nvidia for its cheating by making the 970 a best-seller. If people get upset about corporations cheating their customers they should look in the mirror.
To this day, Nvidia still does not list the proper specifications for the 970 on its website.