AMD Radeon RX 480 PCI-Express Overcurrent Discussion

Discussion in 'Frontpage news' started by Archimonde, Jul 1, 2016.

  1. anticupidon

    anticupidon Ancient Guru

    Messages:
    5,948
    Likes Received:
    2,359
    GPU:
    Polaris/Vega/Navi
    ok, let's see if i got this correctly .
    The card draws more power from the PCI-E port which delivers 75W than the 6 pin power rail.
    If the motherboard is older or a entrance-level, the card will cause power issues, demanding more than PCI-e port can feed it.

    Correct?

    If this is the case, it can be fixed by a revision, or AIB version with 8 pin connector and a tweaked BIOS who limit or adjust to the PCI-E power compliance.
     
    Last edited: Jul 2, 2016
  2. mcfart

    mcfart Master Guru

    Messages:
    303
    Likes Received:
    0
    GPU:
    HD 5970
    Yup. Higher-end mobos have failsafes/"compatibility" for GPUs pulling extra power from the PCIe port, but in older/lower end mobos, it can cause instability/shutdowns/possibly permanent damage. It can't pull more than 75W from the 6-pin, so it relies on the PCIe slot.

    AMD obviously went with high clocks to match a GTX 970, but in essence they made the 480 a 175W product. Yes, they can put an 8-pin, but that will discourage OEMs from putting the 480 in prebuilts since OEM PCs typically have low-end Power supplies and motherboards.

    They should have put a strict 150W power limit on it. Meaning throttling etc (which would likely mean throttling whenever it's playing games, limiting performance). This PR is bad for AMD in regards to getting the 480 into prebuilts.


    Adding failsafes for PCIe cards that go above 75W makes mobos more expensive, so OEMs will avoid the 480 until the issue is resolved.
     
  3. WhiteLightning

    WhiteLightning Don Illuminati Staff Member

    Messages:
    28,916
    Likes Received:
    1,763
    GPU:
    GTX1070 iChillx4
    Official Statement from AMD on the PCI-Express Overcurrent Issue

    Source: http://www.guru3d.com/news-story/amd-gives-statement-on-the-pci-express-overcurrent-problems.html
     
  4. TheDeeGee

    TheDeeGee Ancient Guru

    Messages:
    7,154
    Likes Received:
    1,362
    GPU:
    NVIDIA GTX 1070 8GB
    But what about those who already blew up their motherboard? They arn't mentioning anything about refunds etc etc.
     

  5. alanm

    alanm Ancient Guru

    Messages:
    10,228
    Likes Received:
    2,380
    GPU:
    Asus 2080 Dual OC
    Problem is, whatever fix through software/driver will probably result in less performance for the card. Of course AIBs with 8-pin or dual 6-pins should have no issue at all. I would hope for the reference card they do a quick revision so all future shipments are 8-pin.
     
  6. Grimbarian

    Grimbarian Ancient Guru

    Messages:
    1,760
    Likes Received:
    366
    GPU:
    AMD RX5700 8gb
    I'm sure they'll address those directly on a case by case basis, they'll need proof of purchase of a 480 along with proof of the damage otherwise anyone could try it on for some free money!
     
  7. vase

    vase Ancient Guru

    Messages:
    1,653
    Likes Received:
    2
    GPU:
    -
    Why do they need to mention it?

    So far I can't see any complaints about broken computers.
    One youtuber with a socket AM2 board who showed us how raising power target will shutdown his board. That's all.

    If anyone had damage... Article 2, Section 315 of UCC regulates Implied Warranty. AMD doesn't need to mention anything for commercial laws being in effect. Completely independent of vendor warranty.

    So if you have bought it and it blew up your house. If you're still alive you can write a letter to AMD.
    If you died during the explosion (that must have been very big and may have caught over to neighbor's houses as well, knowing how the 480 operates)...then your next of kin has to take care of that in your place.
     
  8. WhiteLightning

    WhiteLightning Don Illuminati Staff Member

    Messages:
    28,916
    Likes Received:
    1,763
    GPU:
    GTX1070 iChillx4
    Dont think AMD will do anything about the broken hardware.
    You need to contact the seller or board partner not AMD.
     
  9. vase

    vase Ancient Guru

    Messages:
    1,653
    Likes Received:
    2
    GPU:
    -
    sooo... then all the big damage that has already been caused will be shifted over to amd in a separate compensation claim from the add-in-board vendors, i guess?! :( i can't imagine how high those will be following up all the upcoming returns and mainboard RMAs...
    i hope it won't have any governmental consequences. or calls for going into state of emergency.

    my nephew works in a law firm as an assistant and already told me they got about 5 calls per hour yesterday afternoon from angry consumers that wanted to sue AMD for completely vaporized pcie slots. basically it seems the overdraw exactly carves out a hole in the spot where the slot was mounted. its crazy! i will ask him if he can organize some pictures for proof!

    yes that is indeed true, if we wanna be as accurate as possible here.
    and we wan't to i think. because this matter is getting more serious as we speak.
     
  10. eclap

    eclap Banned

    Messages:
    31,495
    Likes Received:
    4
    GPU:
    Palit GR 1080 2000/11000
    Wow, the biggest downgrade story of the year will end up being the 480, not Watch Dogs 2.

    Vase, can you stop taking on the world? It's a graphics card, not sure why it means so much to you. Let it go, seriously, you're acting like it's your first time on the internet.
     

  11. vase

    vase Ancient Guru

    Messages:
    1,653
    Likes Received:
    2
    GPU:
    -
    it is, and now stop ignoring me :nerd:
     
  12. Spets

    Spets Ancient Guru

    Messages:
    3,080
    Likes Received:
    178
    GPU:
    RTX 3090
  13. Turanis

    Turanis Ancient Guru

    Messages:
    1,777
    Likes Received:
    471
    GPU:
    Gigabyte RX500
    Only drops in the ocean,nothing special.

    Any video card(e.g Gtx 750Ti which get 140W) if dont have enough power pins(8,6+8) will get from PCI-e what he want,but not so much to blow the other components.

    On Guru3D bench the RX 480 get ~166W,but spikes not continuous power.
     
  14. Turanis

    Turanis Ancient Guru

    Messages:
    1,777
    Likes Received:
    471
    GPU:
    Gigabyte RX500
  15. Loophole35

    Loophole35 Ancient Guru

    Messages:
    9,793
    Likes Received:
    1,148
    GPU:
    EVGA 1080ti SC
    Techpowerup showed spikes of 300W total power and average continuous draw of 82W from the PCI. Juxtapose that with the 750Ti pulling 68W average continuous with spikes.

    I wonder if this happened because of the switch to GoFlo?
     

  16. Fox2232

    Fox2232 Ancient Guru

    Messages:
    11,810
    Likes Received:
    3,363
    GPU:
    6900XT+AW@240Hz
    Yes, and I have seen quite a few people with 750Ti (60W TDP base) and quite brutal OC without additional power connector.
    There are GTX 950 LP with base 75W and again no power connector.
    And above it there is GTX 960 OEM again no Power connector?
    Radeon R7 360E is 75W base TDP and still no problem?
    Nothing burned?
     
  17. vase

    vase Ancient Guru

    Messages:
    1,653
    Likes Received:
    2
    GPU:
    -
    yes please a review with a test rig that is in the 'budget' segment, so that it fits the very low segment of the card. i suggest components from 2010 or older. no modern motherboard and no modern cpu.

    so we can see how the card works out on the pcs that it was built for.

    and maybe we re-review the majority of all graphic cards every second month to take driver changes into consideration. i heard of up to 10% improvement from some driver changes for certain games.

    a re-review of a card that physically didn't change? really? do you think anything is wrong with how the review was done? let us know.

    you think hilbert hasn't got better things to do than re-review a card to show some undervolted same-clock performance so he just can do the review of the aib cards one week later?
    if undervolted shouldnt reach the before-clocks then we put another power limit on and we have the exact same result as before.
    why would you retest a card that didnt make any problems?


    jeeeesus... so many people are stuck with the failed impression that GPU lifecycle is shorter than the CPU/mainboard one
    or purposely imply it would be like that.
     
  18. Spets

    Spets Ancient Guru

    Messages:
    3,080
    Likes Received:
    178
    GPU:
    RTX 3090
    What are you blubbering on about this time, where in AMD's statement did they say the clock speeds will remain the same and they'll just undervolt the card?
    If the card is going to be limited with this "fix" then the original review of the card doesn't really mean much now does it. It'll also be nice to know if the issue really is fixed instead of remaining ignorant.

    What do you mean the card "didn't make any problems?" losing onboard audio and continually using more power than the PCI-E spec is a problem, one that'll be worse in the long run. Also why would AMD issue a statement and a driver fix for something that isn't a problem?

    Feel free to not reply because I'm pretty tired of reading your nonsense every 2-3 posts in a thread.
     
  19. vase

    vase Ancient Guru

    Messages:
    1,653
    Likes Received:
    2
    GPU:
    -
    oh my bad!

    and where did they say clock speeds will change?

    OH, YOU'R BAD!
     
  20. Spets

    Spets Ancient Guru

    Messages:
    3,080
    Likes Received:
    178
    GPU:
    RTX 3090
    A software fix will clearly throttle the card, I can't see how they can simply undervolt the card and keep the same clocks to fix the PCI-E power draw like you say they will.
    But by all means, keep blubbering. NOW WITH CAPS LOCK ON.
     

Share This Page