RX 480 is apparently killing pcie slots

running Furmark on full screen completely maxxed out @ 1440p, both cards are on stock fan profile with a slight OC @100% usage as we speak. I also threw like 4 youtube play lists on my second monitor. Now we wait.

the furry-butthole of heat will be run till 7:30 PM est until I get home.
 
PCI-E.jpg


Lot of placebo effect floating around. We should've been hearing about dead pci-e slots for years by now.
Spikes vs sustained are two different things.
 
someone tell me what software to run to blow up my computer. Ill remote in from my office install it, and blow it up... Will furmark do fine? ill run that shit for the next 7 hours. Im so happy with my 480's i just bought 2 more to crossfire my wifes PC

Set power limit to 150% on both cards, and run Furmark.

While you're at it, better buy some insurance for your house.
 
you know the issues he was having is probably not new by any means in regards to old cheap amd boards lol....I wouldn't touch one of those ultra cheaply made boards even with nvidia cards.

The boards may be old and cheap, and more sensitive to over-draw of power, but if the rx 480 is drawing more than 75W on the PCIe x16 slot, it is exceeding the specification, so it is at fault, not the motherboard.

There is an argument to be made that the motherboards probably should have a higher safety margin to avoid problems like this, but still, one is complying with the spec, the other isn't.
 
AMD needs to down clock the card ASAP.. Otherwise more casualties will really corrode into AMD's reputation which is already fairly tarnished.

They have a habit of doing last minute clock and voltage adjustments to match the competition, for eg 290/290X bring clocked to the hilt.

Too bad they only put in a 6 pin connector and shot themselves in the arm this time around.

They won't down clock it. What they will do is adjust the p-state which in the end will affect the boost clock to down throttle quicker, which in the end WILL affect performance.

All the reviews out there will be invalidated which is still grounds for deceptive practices and false claims.

Like I said, wait a week. If the "FIX" lowers the performance, and/or locks out overclocking, then AMD better offer to do a refund or RMA to add a 8 pin header. Otherwise they are facing a shitstorm.
 
Last edited by a moderator:
Also, I don't understand why this is the case. I was under the impression that modern GPU's with 12V PCIe connectors draw almost all of their power directly over the connectors from the PSU, not through the motherboard.
 
PCI-E.jpg


Lot of placebo effect floating around. We should've been hearing about dead pci-e slots for years by now.


Please link the orginal article so we can see if its just spikes or sustained power on the strix gtx 960.....
 
This isn't really any surprise. The RX480 is cost cutting in every way.
Heavy synthetized and outsourced IC design to a cheaper region.
Poor noisy cooler.
No DVI, despite its price segment got a heavy DVI base.
PCB/VRM not designed for the usage.
All chips, good or bad used as top bin.
1.15V just to reach the desired clocks.
The sold cards seems NOT to be the one that got PCI-SIG certified.

As I see it one of 3 options will happen.
Software fix as they work on now. Either BIOS or driver. But this could easily mean reduced performance.
Recall/RMA of current cards and removal of the bad chips, both now and future.
Recall and a new chip design is needed. This one can mean Polaris is gone for the rest of the year.
 
Also, I don't understand why this is the case. I was under the impression that modern GPU's with 12V PCIe connectors draw almost all of their power directly over the connectors from the PSU, not through the motherboard.

They attempt to split it, but to primarily draw from the PSU's connectors. There's usually a ratio, like, for every 6W taken from the power connectors, 1W taken from the PCI-E slot. The problem with the reference 480 is that it uses a 6-pin and 50/50 split. At its target 150W, it's right on spec for both PCI-E and ATX. Change the ratio to anything other than 1:1, and it violates one spec to save the other. But since at stock it's hitting 160-165W, it's violating both.

Violating ATX spec is no big deal. Cards have done that for years and PSUs are built to handle it. But the PCI Spec is what concerns me. So I do wish they had used another ratio, maybe 33/67 (55W/110W stock). This also would have given some OC headroom. But, I don't know if that split is technically feasible. I don't think it's as easy as specifying a ratio, but rather, the ratio is derived from the power delivery method. Guess we're all about to get an education on this as this situation unfolds.
 
Violating ATX spec is no big deal. Cards have done that for years and PSUs are built to handle it. But the PCI Spec is what concerns me. So I do wish they had used another ratio, maybe 33/67 (55W/110W stock). This also would have given some OC headroom. But, I don't know if that split is technically feasible. I don't think it's as easy as specifying a ratio, but rather, the ratio is derived from the power delivery method. Guess we're all about to get an education on this as this situation unfolds.


well it definitely is feasible, even the 6990 which uses a ton more power and pulled more from the connectors then it should still didn't pull this much wattage from the PCI-e

I don't know if they can do this without a PCB change though......
 
So I'm guessing the consensus is to just avoid the reference design? This will likely be fixed by the OEM's, as they tend to include more 12v power plugs on their board changes. That should reduce the draw from the motherboard.

PCIe standard is sustained at 75w or less. 6pin PCIe power is rated for sustained 75w, total power continuous is 150w.

From a lot of tests, the RX480 is pulling 160-170w and upwards of 200w when over clocked. That'd end up pushing both the PCIe bus and PCIe 12v past their rated standards.

Should have just slapped an 8pin on there.
 
My cards been running a whole 2 days and hasn't even caught fire. I suspect this wont be a problem for the vast majority of users, over pci-e usage has been shown in various models over the years and casues little in the way of issues. It was fairly common with AGP too.
 
well it definitely is feasible, even the 6990 which uses a ton more power and pulled more from the connectors then it should still didn't pull this much wattage from the PCI-e

I don't know if they can do this without a PCB change though......

They likely all tie into the same power plane. But additional circuitry would be needed on the PCB to prevent back currents. Only a probe to the VRM's can determine this though.

*Note: putting them all in the same power plane MEANS that you expect an equal amount of pull from each power source. This however would simplify design and save cost.
 
Last edited by a moderator:
My cards been running a whole 2 days and hasn't even caught fire. I suspect this wont be a problem for the vast majority of users, over pci-e usage has been shown in various models over the years and casues little in the way of issues. It was fairly common with AGP too.

Even 5% of users having burnt out boards or crashing is serious. That's like saying, "Defective airbags only killed .1% of people in accidents" Look what that cost Takata.
 
So the first graph was a slight OC to 1300 MHz and the power limit set to 150%. I get that, because you may not have been able to keep the OC without the PL increase. Why is the power limit still at 150% when testing at stock frequency with furmark? Can you vouch for whether or not this card acts this way when the power limit is set to default?

AMD Overdrive warning
WARNING: AMD and ATI processors are intended to be operated only within their associated specifications and factory settings. Operating your AMD or ATI processor outside of official AMD or ATI specifications or outside of factory settings, including but not limited to the conducting of overclocking (including use of any overclocking guide or software, even if such documentation or software has been directly or indirectly provided by AMD or otherwise affiliated in any way with AMD), may damage your processor and/or lead to other problems, including but not limited to, damage to your system components (including your motherboard and components thereon (e.g. memory)); system instabilities (e.g. data loss and corrupted images); reduction in system performance; shortened processor, system component and/or system life; and in extreme cases, total unrecoverable system failure. AMD does not provide support or service for issues or damages related to use of an AMD or ATI processor outside of official AMD or ATI specifications or outside of factory settings. You may also not receive support or service from your board or system manufacturer. Please make sure you have saved all important data before using this overclocking software. DAMAGES CAUSED BY USE OF YOUR AMD OR ATI PROCESSOR OUTSIDE OF OFFICIAL AMD OR ATI SPECIFICATIONS OR OUTSIDE OF FACTORY SETTINGS ARE NOT COVERED UNDER ANY AMD PRODUCT WARRANTY AND MAY NOT BE COVERED BY YOUR BOARD OR SYSTEM MANUFACTURER’S WARRANTY. SEE THE AMD OVERDRIVE UTILITY END USER LICENSE AGREEMENT OR THE APPLICABLE AMD OVERCLOCKING UTILITY FOR DETAILS


You can conclude AMD allows you freedom to blow up your own system if you want - no holding hands in other words. (y)
 
Also, I don't understand why this is the case. I was under the impression that modern GPU's with 12V PCIe connectors draw almost all of their power directly over the connectors from the PSU, not through the motherboard.

the problem is not the PCI-E 6 pin power connector, it's the Motherboard PCI-E SLOT, the card it's out of balance, any extra power needed on that GPU it's requested through the PCI-E SLOT power circuitry instead of being request through the PCI-E 6 pin power connector

The boards may be old and cheap, and more sensitive to over-draw of power, but if the rx 480 is drawing more than 75W on the PCIe x16 slot, it is exceeding the specification, so it is at fault, not the motherboard.

There is an argument to be made that the motherboards probably should have a higher safety margin to avoid problems like this, but still, one is complying with the spec, the other isn't.

indeed some pretty good and high quality motherboards have extra power connectors to deliver more power and dedicated current to the PCI-E Slots, also are made with extra PCB layers to ensure better reliability and safety, etc.. however as was said before in this thread and I said since this whole mess started, this isn't the card targeted to people with Higher-end motherboards. this ins't the card targeted to people with higher quality of PSUs, this is the card aimed for those guys with the cheaper possible motherboard with a single PCI-E slot to use GPU you know those one with barely a +4 pin power for the CPU, with a 20$-30$ case with bad airflow, which also buy the cheaper PSU with a good name at best buy (if its thermaltake its good, if its cooler master its good, even if its corsair its good, specially this later which have extreme crap PSUs at their lower end categories). so what you got?. a lot of potential failures certainly specially with a GPU like this one can cause.
 
AMD Overdrive warning


You can conclude AMD allows you freedom to blow up your own system if you want - no holding hands in other words. (y)

Yes but AMD is already running outside spec STOCK. And to offer overclock on top of that is NOT reasonable at all when you are already outside spec. That's down right negligent.
 
This is kinda bullshit. I feel like every time some new thing comes out it turns into a fight to belittle that item... This feels like this particular issue is on the next level and its Salem witch hunting levels. You guys are going ape shit over some possibly bad cards or mobos or even psus/eletricity, It will take some time before this is ever addressed as even a issue.
 
This is kinda bullshit. I feel like every time some new thing comes out it turns into a fight to belittle that item... This feels like this particular issue is on the next level and its Salem witch hunting levels. You guys are going ape shit over some possibly bad cards or mobos or even psus/eletricity, It will take some time before this is ever addressed as even a issue.

A week is reasonable to reply. They have 6 days left.

If I get a service report, I'm on it that day and usually have the fix and analysis in 24 hours. In fact I usually stay up late that night if it's serious enough.
 
They attempt to split it, but to primarily draw from the PSU's connectors. There's usually a ratio, like, for every 6W taken from the power connectors, 1W taken from the PCI-E slot. The problem with the reference 480 is that it uses a 6-pin and 50/50 split. At its target 150W, it's right on spec for both PCI-E and ATX. Change the ratio to anything other than 1:1, and it violates one spec to save the other. But since at stock it's hitting 160-165W, it's violating both.

Violating ATX spec is no big deal. Cards have done that for years and PSUs are built to handle it. But the PCI Spec is what concerns me. So I do wish they had used another ratio, maybe 33/67 (55W/110W stock). This also would have given some OC headroom. But, I don't know if that split is technically feasible. I don't think it's as easy as specifying a ratio, but rather, the ratio is derived from the power delivery method. Guess we're all about to get an education on this as this situation unfolds.

Yeah, I would argue that if you had to violate one of the two, you are much less likely to run into trouble when over-drawing from a PCIe 12V connector, than over-drawing from the PCIe slot.
 
This is kinda bullshit. I feel like every time some new thing comes out it turns into a fight to belittle that item... This feels like this particular issue is on the next level and its Salem witch hunting levels. You guys are going ape shit over some possibly bad cards or mobos or even psus/eletricity, It will take some time before this is ever addressed as even a issue.
Hey guys, it's a unbeliever!!! Get him!!! See if he floats!!!
 
If that Rampage owner with 3x480s gets denied the RMA, prepare for a shitstorm.


The sound chips on the rampage dies pretty frequently. The onboard sound on my Rampage died too during yearly maintenance of my loop. I'm not even sure how, it just makes static and the front ports are out just like that guy from ocn. Did mine die from too much power too? This happened a year ago. Double checks, yep RIVE just like mine.
 
This is kinda bullshit. I feel like every time some new thing comes out it turns into a fight to belittle that item...
My memory may not be perfect, but I don't recall much belittling of the GTX 1080. Just a lot of complaining from people who wanted one that NVidia was screwing them with an early adopters tax (pretty much true, but get used to it, folks) and that there wasn't enough supply to meet the (apparenlty ferocious and not-very-price-sensitive) demand.
 
Also, I don't understand why this is the case. I was under the impression that modern GPU's with 12V PCIe connectors draw almost all of their power directly over the connectors from the PSU, not through the motherboard.
You are correct, this is what AMD should have done.
But they only fitted a 6 pin power connector which is specced for 75W.
So it has to get the rest of the load from the motherboards PCIE-e power connections giving a max, in spec, total power delivery of 150W.
(The mobo is only specced for 75W max)
They are exceeding this on both power connections.

If they had fitted an 8 pin connector they could have had up to 225W and been comfortably home and dry.
Stupid of them.
 
You are correct, this is what AMD should have done.
But they only fitted a 6 pin power connector which is specced for 75W.
So it has to get the rest of the load from the motherboards PCIE-e power connections giving a max, in spec, total power delivery of 150W.
(The mobo is only specced for 75W max)
They are exceeding this on both power connections.

If they had fitted an 8 pin connector they could have had up to 225W and been comfortably home and dry.
Stupid of them.

Yeah, I don't understand how something as simple as this makes it out of the design phase. An 8-pin connector would have costed - what - 2 cents more per video card?
 
Well, it looks like Kyle was right:
In the simplest terms AMD has created a product that runs hotter and slower than its competition's new architecture by a potentially significant margin.

As more and more review sites are testing the power draw of the cards it looks as though there is a problem. PCPer showed that their card is drawing over 75W from PCIe. At stock they showed that most of the time the card is using around 150W total. This indicates that there is no power margin in the design of the card.

What must have happened (and Kyle's editorial points to this) is that they designed the cards to run at either lower clock speeds, or expected the chips to run at the current stock clocks while using less power. When they got the chips in they probably weren't getting the performance they needed at lower clocks to be competitive (which would have given them some power margin), so they decided to effectively factory overclock the chips to get the performance they wanted. Most of the reviewers had trouble overclocking their cards so that seems to indicate that the stock clocks are overclocked from where the design wanted to be.

It would be interesting to see what the power numbers would look like if someone down-clocked a card by even as little as 5-10%. Increasing the clocks by small percentages dramatically increased power consumption for most reviewers. I'm guessing down clocking the card would improve power draw quite a bit.

They should have just designed in an 8-pin power connector. And I was thinking about getting one of these cards. I may still get one of the custom cards if they have a better power delivery setup.
 
Yes but AMD is already running outside spec STOCK. And to offer overclock on top of that is NOT reasonable at all when you are already outside spec. That's down right negligent.
I understand what you are saying, which has already been stated here more then once - it should have had an eight pin connector and power never exceeding 75w ever from the pcie slot - overspec power use always comes from the pcie power connector. I am sure AMD now has lawyers involved since it is more of a Pandora's box.

My suggestion is a total recall with option of you keeping the card knowing the concerns :whistle:. AMD allowing you freedom to choose so to speak. All those cards, good grief will need to be retrofitted with an eight pin connector (supposedly traces on the board already). I hate the idea of loosing OverDrive though big time but that is another option if these cards will pass at stock settings.
 
This is kinda bullshit. I feel like every time some new thing comes out it turns into a fight to belittle that item... This feels like this particular issue is on the next level and its Salem witch hunting levels. You guys are going ape shit over some possibly bad cards or mobos or even psus/eletricity, It will take some time before this is ever addressed as even a issue.

The card draws more power from the motherboard than is allowable by spec. And, it seems to impact the majority of cards. Every card tested at the PCI-E connector has shown this behavior, both review and retail samples. Show me ONE review that tested power at the PCI-E connector using a riser card that doesn't show PCI violation.

It's not a witch hunt. It's damage control by people who don't believe this.
 
Yeah, I would argue that if you had to violate one of the two, you are much less likely to run into trouble when over-drawing from a PCIe 12V connector, than over-drawing from the PCIe slot.

So, Who here is good with board design. Is something like this likely to be tweakable after the fact? Could they address where the power is pulled from using a firmware update to shift more power to the PCIe 12v connector, and less to the PCIe slot? Or is it more likely that they are going to have to recall these early boards?

God knows AMD can't afford to be spending their limited cash to fix a mess like this right now. I'm certainly hoping this can be affordably fixed.
 
So, Who here is good with board design. Is something like this likely to be tweakable after the fact? Could they address where the power is pulled from using a firmware update to shift more power to the PCIe 12v connector, and less to the PCIe slot? Or is it more likely that they are going to have to recall these early boards?

God knows AMD can't afford to be spending their limited cash to fix a mess like this right now. I'm certainly hoping this can be affordably fixed.

We dont know, it could be as simple as a bios update, it could require a recall, we really don't know at this point
 
So, Who here is good with board design. Is something like this likely to be tweakable after the fact? Could they address where the power is pulled from using a firmware update to shift more power to the PCIe 12v connector, and less to the PCIe slot? Or is it more likely that they are going to have to recall these early boards?
Who cares!!! They're guilty! Get them!!!
 
Back
Top