Welcome Stranger to OCC!Login | Register

Newly Released RX 480 Appears to Have Power Draw Issues

Category: Video Cards
Posted: 01:38PM
Author:

Just days ago, AMD released the long-awaited RX 480 graphics card, containing the new Polaris 10 GPU. One of the features of the new GPU is supposed to be significantly increased efficiency, thanks to the new 14 nm FinFET node it uses and several Radeon technologies, but some reviewers have found the card, at least in the currently available reference design, may have significant power issues.

The card has a TDP of 150 W and a 6-pin power connector, which is meant to provide 75 W, leaving the other 75 W to come from the PCIe slot itself, which is the maximum the slot is supposed to provide. What some reviewers have found, including those at PC Perspective who have done a careful analysis of the card's power draw, is that the RX 480 is pulling above 150 W, at stock (overclocking will naturally draw even more power). On its own, pulling more power than the stated TDP is not necessarily an issue, especially if it draws it from the PSU connector, but the PCPer reviewer found it was also drawing more than 75 W from the PCIe slot. Depending on the tolerances of the motherboard, continuous power draw above the spec could cause damage to the slot. Obviously, this is not a good.

As some comments for the original PCPer article suggested examining the ASUS GTX 960 Strix, which was supposedly caught also drawing more from the PCIe slot in the past, the review pulled out his testing rig again and got to work. He found that this 960 does not in fact exceed that limit, at least not continuously. The reviewer used a second order low pass filter specifically to remove some of the spiking that can occur with phase switches of any DC switching power supply. This makes the data easier to read and the reviewers focus was on continuous power draw, not instantaneous.

It is also worth noting that there has not been any official response from AMD concerning this issue yet. The most I have seen has been an Reddit AMA response by Raja Koduri himself stating the RX 480 passed testing and that the engineering team is fully engaged on this matter. So for now, the exact nature, cause, and extent of this issue are not known, at least not publicly, and neither is if a potential fix would be software or hardware based.

Source: PC Perspective



Register as a member to subscribe comments.
N.E.A on July 01, 2016 23:27
Why is it that manufacturers can't get a card out right without issues with the most basic functions. If it tells me something, it is that they are all after quick money. If they had taken their time to make it right, they'd make a lot more.
ir_cow on July 02, 2016 04:53

They could have fixed this with an 8-Pin instead of using 6Pin. I guess they really wanted people to think it had a low power draw.

DrTiCool on July 02, 2016 09:13
AMD simply screwed up this time, power design wise. Im some scenarios card draws almost 200W. I expected this new 14nm FinFet shit act like or even better than Nvidia 16nm stuff, performance per watt. Not happening. GloFo process sucks compared to TSMC.
Guest_Jim_* on July 02, 2016 13:16

AMD has started sending out official statements now (from TechPowerUp):

"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 8 Gbps 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)."

 

Software fix incoming for the apparently incorrect tuning on some cards. More information will be provided on Tuesday, if not the driver update itself.

By the way, I've been seeing reports of people successfully undervolting their RX 480s without having to lower clocks (for some they could actually overclock better because the heat was kept down).

Braegnok on July 02, 2016 13:55

They could have fixed this with an 8-Pin instead of using 6Pin. I guess they really wanted people to think it had a low power draw.

 

+1,.. the 8-pin would have been the best way too provide 150W and just use the PCIe lane for headroom/overclocking.

 

Wouldn't want to be the guy who said go with the 6-pin after seeing the power draw during RD, :slap:   

Guest_Jim_* on July 02, 2016 14:15

 

They could have fixed this with an 8-Pin instead of using 6Pin. I guess they really wanted people to think it had a low power draw.

 

+1,.. the 8-pin would have been the best way too provide 150V and just use the PCIe lane for headroom/overclocking.

 

Wouldn't want to be the guy who said go with the 6-pin after seeing the power draw during RD, :slap:   

 

Except this level of power draw might not be intentional or known by AMD prior to release. It is possible there is a fluke of situational influences causing the BIOS to request higher volts and more power erroneously or that some cards were flashed with an incorrect BIOS that consistently draws more power than needed.

I'm not saying an 8-pin wouldn't have been a good idea (though I could see the 6-pin being intentionally picked so that the AIB manufacturers can have the additional headroom of other power configurations) but it is also possible this issue did not present itself during the RD and QA phases of development. Until more information is provided on the issue (Tuesday) or the fix goes live and people start searching for the changes, we don't know what the cause is.

jdm_freek on July 03, 2016 20:23

dont all cards do this on occasion when overclocked??????????

and why does it matter so much.

Waco on July 04, 2016 23:20

dont all cards do this on occasion when overclocked??????????
and why does it matter so much.

Overclocking does draw more, yes, but generally that power is drawn from the PCIe cables, not your motherboard. Pulling above rated on the PCIe cables isn't a huge deal...asking your board to do that can quickly cause permanent damage.
jdm_freek on July 05, 2016 19:35

but this is only effecting older boards that have poor power allocation 

im having no issues what so ever.

SpikeSoprano on July 06, 2016 02:11

but this is only effecting older boards that have poor power allocation 
im having no issues what so ever.


The problem is not older mb's, the problem is the card draws over 50 watts more than amd said it would , if they had stated that in all their hype it could fry your mb I doubt they would sell very many.
Braegnok on July 06, 2016 03:33

Overclocking new high-end motherboards will cause heat issues just as fast as older mid-range motherboards. The only difference is the high-end boards will handle more heat/volts longer before failing.

 

I was cleaning my workstation last week and noticed it's been getting a bit toasty.  [attachment=20915:20160705_205820.jpg] 

Waco on July 06, 2016 12:27

but this is only effecting older boards that have poor power allocation
im having no issues what so ever.

Killing your motherboard should absolutely concern you. I wouldn't overclock at all till they sort this out!
Guest_Jim_* on July 06, 2016 12:34

AMD has released the promised statement and will be releasing a new driver within the next 48 hours (so Thursday). This new driver is going to shift more of the power draw onto the 6-pin connector, thereby reducing the draw from the PCIe slot. (The 6-pin connector is more resilient to drawing more power than it is spec'ed for, so a modern PSU shouldn't have a problem. Also I've seen some claims that because of how the power connector is designed in the RX 480, it can treat it like an 8-pin to draw more power. Something about the 6-pin configuration normally having one unpowered, but the card can have that pin be powered. I don't know how true this is though, hence the parentheses.) It will also add a Compatibility Mode toggle, that when enabled (default is off) will apparently drop the voltage of the card, to bring it under the 150 W. The performance impact from this AMD has said should be minimal, and as some people who have been testing with undervolting have found the performance improves, thanks to reduced temperatures, AMD's claim is very possibly accurate, but it will not be until tests are done with the new drivers that we can know for sure. Also the drivers have some new optimizations that should increase performance by ~3%, which could offset some of the Compatibility Mode loss. Here's the Anandtech article on it: AMD's Tuesday Radeon RX 480 Status Update: New Driver By Late Thursday.

 

By the way, SpikeSoprano, where were you seeing the 50 W over TDP figure, because the only time I've seen reports of it crossing 200 W was when people overclocked the card? That's not to say the times stock reached into the 160s and 170s W are not issues, but those are still a lot closer to 150 W than 200 W.

Also, I'm on vacation currently, so I do not have ready access to the Internet to post more, but I wanted to make sure this got up here. Back to vacationing and not being on the Internet.

Braegnok on July 07, 2016 21:17
SpikeSoprano on July 08, 2016 03:08

By the way, SpikeSoprano, where were you seeing the 50 W over TDP figure, because the only time I've seen reports of it crossing 200 W was when people overclocked the card? That's not to say the times stock reached into the 160s and 170s W are not issues, but those are still a lot closer to 150 W than 200 W.
Also, I'm on vacation currently, so I do not have ready access to the Internet to post more, but I wanted to make sure this got up here. Back to vacationing and not being on the Internet.

That is what I meant, on overclocked cards, which is the first thing a lot of people do when they get a new card .
ps> hope you enjoy your vacation !

This news has comment postings disabled because it is now archived.

© 2001-2017 Overclockers Club ® Privacy Policy
Elapsed: 0.2056608200   (xlweb1)