AMD Updates Its Users About RX 480 Power Issues

Subject: Graphics Cards | July 6, 2016 - 08:11 PM |
Tagged: rx 480, Polaris, graphics drivers, amd

In the next 24 hours or so, AMD will publish Radeon Software 16.7.1, which addresses the power distribution issues in the AMD Radeon RX 480. The driver makes two major changes. First, AMD claims that it will lower the draw from the PCIe bus. While they don't explicitly say how, it sounds like it will increase the load on the 6-pin PCIe cable, which is typically over-provisioned. In fact, many power supplies have 6-pin connectors that have the extra two pins of an 8-pin connector hanging off of it.

View Full Size

Second, seemingly for those who aren't comfortable with the extra load on the 6-pin PCIe connector, a UI control has been added to lower overall power. Being that the option's called “compatibility”, it sounds like it should put the RX 480 back into spec on both slot and the extra power connector. Again, AMD says that they believe it's not necessary, and it seems to be true, because that option is off by default.

View Full Size

Beyond these changes, the driver also adds a bunch of game optimizations. Allyn and Ryan have been working on this coverage, so expect more content from them in the very near future.


July 6, 2016 | 08:31 PM - Posted by pdjblum

Allyn and Ryan could have taken the few minutes it would have taken to post this in the am as every other site did, especially since they were the ones that milked this for all it was worth. They had an obligation to be forthcoming. Gross negligence and unethical intent are pretty much equally harmful. Certainly, AMD was accused of both in this case. Thanks for doing the right thing and posting this Scott.

July 6, 2016 | 09:52 PM - Posted by Anonymous (not verified)

Oh come on... that's just a sad excuse to hammer them for something.

Here, if you think they were negligent in posting this story, show how busy they were today.

July 7, 2016 | 12:06 AM - Posted by Anonymous (not verified)

You are too hurt, maybe you should be on the green team, because AMD does not need your kind of illogical adoration. Ryan was doing a good and proper amount of work on his analysis of the problem. The motherboard makers(The Good ones) have been building their designs knowing that the GPU makers have been right on the edge of PCI compliance over the years! So most Motherboards are built to handle some amounts of outside the specifications behavior from any new GPU products to assure that there are very few motherboard RMA's needed on these makers' motherboard SKUs. AMD has a fix incoming, and all GPU makers need to have their products properly analyzed for any issues no matter how small.

July 7, 2016 | 02:50 AM - Posted by JohnGR

The motherboard guys build also RX 480 cards, ASUS and MSI for example. Would they produce a card that would burn their own boards? They would had rejected AMD's design.

July 7, 2016 | 02:47 AM - Posted by JohnGR

When I read that Raja was going to give an exclusive to PCPer about the RX 480 my first thought was

"This guy is naive or stupid".

July 6, 2016 | 09:15 PM - Posted by StephanS

I recall a long time ago some website would track ocerclocking and voltage requirements.

I wonder how many RX 480 work fine with a 10% voltage drop,
because so far it seem to give better performance (because it draw less power and generate less heat == higher sustain clock)

PCPer... please explore this with your sample card ?

(Its not all about overclocking anymore to get better performance.)

July 6, 2016 | 09:26 PM - Posted by quest4glory

It's not necessary, unless you're that one guy who plugs it into the PCIe slot of an OEM motherboard from a 2007 Compaq.

July 6, 2016 | 09:51 PM - Posted by Anonymous (not verified)

Or you know, some AMD specific boards that are older than sin. Well, okay not sin. 990FX was Q2 2011 (So five years). Over time + cheaper boards (which like or not that is what AMD is at the moment) is a recipe for disaster.

July 6, 2016 | 09:50 PM - Posted by djotter

This guy has done what the driver will likely do to redistribute the load to favour the 6-pin.

http://www.overclock.net/t/1604979/a-temporary-fix-for-the-excess-pci-e-...

July 7, 2016 | 12:26 AM - Posted by donut (not verified)

If you can't wait just undervolt. And gain performance lol.

http://www.legitreviews.com/amd-radeon-rx-480-undervolting-performance_1...

July 7, 2016 | 02:42 AM - Posted by JohnGR

I am still waiting for PCPer to get a GTX 950 with NO power connector, overclock it and see how much power it will draw from the pcie bus.

At Techpowerup, in their review the card was drawing 74W on typical, 76W maximum, and 79W peak at DEFAULTS, when gaming. After overclocking the card, they saw a 20% increase in PERFORMANCE.

Performance does NOT come as free gift. I am pretty sure in this case 20% extra performance means at LEAST 20% extra power consumption, which means 90W from the pcie bus.

This doesn't change anything about RX 480 that consumes more it should at DEFAULTS. But it shows that MANY cards do that after they are overclocked.

But OK, this is PCPerspective. PCPER WILL NEVER POST SOMETHING AGAINST NVIDIA IF IT IS NOT ALREADY CIRCULATING ALL OVER THE INTERNET.

This is techpowerup's review
https://www.techpowerup.com/reviews/ASUS/GTX_950/

power consumption at DEFAULTS
https://www.techpowerup.com/reviews/ASUS/GTX_950/21.html

unfortunately they DIDN'T measure power consumption under overclocking, but 20% extra performance guaranties that the card DOESN'T throttle to maintain that 75W limit
https://www.techpowerup.com/reviews/ASUS/GTX_950/26.html

One last thing. The GTX 960 that it was used as comparison to RX 480 comes with a much lower than 150W TDP and an extra 6pin power connector. I really don't know which moron started the whole fuss about that card going over 75W when in fact it had plenty of power connectors on it. Was it Nvidia's marketing department?

July 7, 2016 | 07:55 AM - Posted by Anon (not verified)

Seriously, who cares. The 960 without a power connector wasn't a common card and wasn't the reference design. Not to mention it's old now.

It's pretty childish to go on pointing the finger at others when encountered with an issue.

July 7, 2016 | 04:27 AM - Posted by Anonymous (not verified)

AMD fanboys are the most unreasonably defensive people on earth. Always playing the victim card.

July 7, 2016 | 07:12 AM - Posted by Anonymous (not verified)

Wow, PCper put out a 100% fact based, unbiased and detailed review of an issue that obviously existed, AMD would never have made changes if it didn't, and the massive troll fan base for AMD still can't accept the fact that AMD put out a card that was dangerous.

I have nothing against AMD, and recommended this specific card heavily before its release based on information available at the time. But the level of stupidity and willful ignorance by the people who refuse to accept how serious of an issue this was is astounding.

Sadly AMD, even though they are trying to address the issue, still have not done the right thing. They are releasing a forced fix, and giving people a tool to limit the power draw even more. While in the same breath pretending as if it was never an issue in the first place.

There is a reason Intel and Nvidia are kicking the snot out of AMD, and no it's not because over 80% of PC gamers are "stupid fanboys". It's because AMD puts out inferior products, and is less trustworthy than Bill Cosby at an open bar. The put out a card that bricked customer's systems, and straight up ripped people off with the whole 4GB to 8GB nonsense. Everyone who paid $239 for an 8GB model is owed a $40 refund.

July 7, 2016 | 09:01 AM - Posted by donut (not verified)

Wanna tissue?

They say they've got a fix for it so lets give them a chance. And just relax and smell the roses man.

The custom 480's should be good if the price is right. I'm definetly interested. My last two Amd cards have been good.

July 7, 2016 | 09:08 AM - Posted by Anonymous (not verified)

http://www.tomshardware.com/reviews/amd-radeon-rx-480-power-measurements...

Considering the customer systems bricking power issue. Also there are multiple reviews where the RX 480 reference model has been overclocked and it has not "bricked" the reviewers test system.

When it comes to the 4gb models being possible to flash to 8gb, doing so voids the warranty and since it has been configured to be a 4gb model, I see no problem with them selling the cards as 4gb models. Never seen anyone complain about cards that could have been overclocked to be equilevant with higher cost aftermarket cards either, why is this such a big problem?

July 9, 2016 | 05:25 PM - Posted by Kusanagi (not verified)

How many systems were "bricked"? it likely wasn't very many. most modern motherboards have very good phase variance, So they're capable of handling some extra heat. the issue would really only impact users of older/lower end boards that don't have very good phase variance.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.