Slashdot is powered by your submissions, so send in your scoop

 



Forgot your password?
typodupeerror
×
Graphics Games

Amazon MMO New World Is Bricking RTX 3090s, Players Say; Amazon Responds (gamespot.com) 144

An anonymous reader quotes a report from GameSpot: Amazon [...] is now bricking high-end graphics cards with a beta for its MMO, New World, according to players. Amazon has now responded to downplay the incident but says it plans to implement a frame rate cap on the game's menus. According to users on Twitter and Reddit, New World has been frying extremely high-end graphics cards, namely Nvidia's RTX 3090. It's worth noting that while the RTX 3090 has an MSRP of $1,500, it's often selling for much more due to scarcity and scalpers, so players could easily be losing upwards of $2,000 if their card stops working.

Specifically, it seems that one model of the RTX 3090 is being consistently fried by New World. On Reddit, a lengthy thread of over 600 posts includes multiple users claiming that their EVGA 3090 graphics cards are now little more than expensive paperweights after playing the New World beta. The "red light of death," an indicator that something is disastrously wrong with your EVGA 3090, doesn't pop up consistently for players though. Some report their screen going black after a cutscene in the game while others have said that simply using the brightness calibration screen was enough to brick their card.
Amazon Games says a patch is on the way to prevent further issues. "Hundreds of thousands of people played in the New World Closed Beta yesterday, with millions of total hours played. We've received a few reports of players using high-performance graphics cards experiencing hardware failure when playing New World," said Amazon Games in an official statement.

"New World makes standard DirectX calls as provided by the Windows API. We have seen no indication of widespread issues with 3090s, either in the beta or during our many months of alpha testing. The New World Closed Beta is safe to play. In order to further reassure players, we will implement a patch today that caps frames per second on our menu screen. We're grateful for the support New World is receiving from players around the world, and will keep listening to their feedback throughout Beta and beyond."

New World is currently set to launch for PC on August 31.
This discussion has been archived. No new comments can be posted.

Amazon MMO New World Is Bricking RTX 3090s, Players Say; Amazon Responds

Comments Filter:
  • StarCraft II (Score:5, Insightful)

    by fazig ( 2909523 ) on Saturday July 24, 2021 @09:12AM (#61615369)
    Bricked two of my gForce 8800GT or what they were called back then, because the 2D elements of the GUI rendered at such a high rate that it caused overheating.

    Some part of the blame certainly goes to Blizzard, like it does to Amazon here. But most of the blame is on nVidia and or the AIB partner for not implementing proper safeguards to prevent this from happening. If such high temperatures are detected by the sensors, they ought to throttle at which rate the affected component is clocked in order to prevent damage.
    So I presume in order to not have the hardware throttle nVidia assumes that what gaming developers do is always good and let them get around the safeguards that should protect the hardware.

    I mean I run tools like Furmark to establish stability of a GPU. And that one is explicitly designed to be a power virus. Yet so far it has never killed any GPU that I've tested with it (yes, not even those old 8800GT).
    • Re:StarCraft II (Score:5, Insightful)

      by AmiMoJo ( 196126 ) on Saturday July 24, 2021 @10:24AM (#61615571) Homepage Journal

      The 3090 does thermally throttle to protect itself.

      My guess would be that the 2D rendering here is stressing one small part of the GPU that doesn't have a temperature sensor nearby because Nvidia figured it's never going to get hammered.

      • by fazig ( 2909523 )
        I have to admit that I don't have the details of those cases.
        All I can say is that it reminds me of that StarCraft II issue, where you had to force the game into VSync to prevent it from happening until Blizzard fixed the issue from their side. It would be somewhat embarrassing if it turns out to be the same or a very similar issue after over 10 years.

        Maybe it was the users that screwed with something they shouldn't have. Some do use custom firmware in order to be better able to overclock their graphics
      • But note the story is about PLAYERS blaming the game maker. This is a good example of ignorance (not intentional) causing a problem.

      • by MrL0G1C ( 867445 )

        I've seen menus run at thousands of FPS many times in different games that didn't cap the FPS, Nvidia should know that it's a common scenario.

        Amazon are a bunch of c**ts for not immediately halting the game until the patch is fully rolled out. Sure this isn't exclusively their fault but continuing to brick $2000 cards because it only affects some people is complete shithead move. Anyone who's card is bricked by the game after Amazon knew about the issue should probably sue them* *I am not a lawyer, you have

      • i don't see amazon at fault here. if the rendering is simple enough for a card to display it at huge fps, it is up to the card to throttle itself. EVGA is the only 3090 card bricking, so it's clearly an issue on their side

    • Some part of the blame certainly goes to Blizzard, like it does to Amazon here.

      No it doesn't. It's up to the hardware and firmware to be designed not to fail when standard system calls are made to it. If some non-standard calls are made tweak firmware / driver settings and bypass safety mechanisms such as power limits and thermal throttling then *and only then* does some Blame go to the programmer.

      Otherwise the blame lies exclusively and exhaustively on the vendor.

    • Re:StarCraft II (Score:5, Insightful)

      by r2kordmaa ( 1163933 ) on Saturday July 24, 2021 @10:54AM (#61615677)
      Absolutely no blame goes on Amazon here. It's the most basic assumption in all of software development, that no matter what you write, the compute hardware will not literally halt and catch fire on you. You should not be able to do so even deliberately - because of course some malware developer will try if it's possible. Compute hardware must be built to be inherently safe from such bugs and if it's not it's entirely on the shoulders of hardware manufacturer and nobody else.
    • I know, I floored my car in neutral for a while and the engine blew up, so I blamed the guy who sold me the gas for not warning me about that.

      He said that $CAR_MAKER should have put some kind of RPM limiter doohickey in there but what does he know?

      • by rtb61 ( 674572 )

        This makes no sense. Timing issues are all built in, any part of the electronic device, should only clock as fast as it was designed to do. No software should be able to alter that, unless it alters the firmware controlling timing cycles. It all has to work together, to be timed to work together as controlled by the firmware, to not clock any faster than set by the firmware. Software should not make a difference.

        Unless there is actual evidence of hardware failure (actual burnt out component), then likely th

    • by fermion ( 181285 )
      Yeah, my car has a rev limiter and a heat sensor that keeps me from killing the engine. All the blame goes to the card manufacturer and they should replace them. Or people should stop buying defective products because they come in a big expensive box.
    • by Kisai ( 213879 )

      Yes and no.

      While nvidia and amd should be blamed for permitting the GPU to achieve these temperatures in the VBIOS, generally the highest end models are a problem because usually they are not performance capped like their weaker siblings.

      In the case of "the menu" triggering it, it's because the user has turned VSync off, and/or has a high-refresh rate monitor (eg 144, 240, etc) so the GPU tries to maximize the framerate. Getting 10,000fps on the menu is literately a spin-wait problem where the game should h

      • by fazig ( 2909523 )
        It certainly wouldn't be the first time EVGA gets some attention for their high end models in the RTX 3000 series.
        Last year we had a bit of a controversy around EVGA's choice of capacitors on their cards, which could lead to crashes and other instability issues.
      • by fazig ( 2909523 )
        Igor'sLAB did some digging here https://www.youtube.com/watch?... [youtube.com] (unfortunately only available in German). But he also writes articles that get translated: https://www.igorslab.de/en/evg... [igorslab.de]

        tl;dr: In EVGA's case, alegedly it's the fan control leading to eventual failure of the GPU fan. Though it's still weird that this would fry the GPU instead of EVGA or nvidia putting in some mechanism that shuts the thing off if it gets too hit.

        From personal experience I'm not sure what to do with that information.
  • by Lisandro ( 799651 ) on Saturday July 24, 2021 @09:14AM (#61615377)

    Sounds like a typical factory-overclocked "gamer" card going out of thermal specs.

    • by Fringe ( 6096 )
      I understand your assumption, but years ago, around ten or so, a new CAD release melted my (mid/high-range) graphics card. It turned out to be a rare occurrence, only a few models of cards were affected, and not all of them, probably some combination of CPU, card and settings. But absolutely on stock unmodified equipment. Lightning sometimes strikes people who aren't holding lightning rods.
      • by Lisandro ( 799651 ) on Saturday July 24, 2021 @09:28AM (#61615399)

        I understand: that's still a hardware failure. GPUs are not suppsed to overheat and die even if you run them full tilt.

      • I understand your assumption, but years ago, around ten or so, a new CAD release melted my (mid/high-range) graphics card.

        No it didn't. Your card not having correct firmware limits set by the OEM melted your graphics card. Your CAD release simply gave it a workload it should have taken gracefully.

        Blame where blame is due, unless your software tweaked hardware settings or bypassed safety mechanisms then blame lies squarely on the vendor of the hardware.

        If I sell you a car with a tachometer showing the redline at 12000rpm, and a rev limiter set at 12000rpm, then it is definitely NOT your fault if the car blows up while you're dr

      • I suppose that was the same class of "rare occurrence" as the FDIV bug?
  • by Jabes ( 238775 ) on Saturday July 24, 2021 @09:25AM (#61615391) Homepage

    If it is true that this software is bricking the graphics card, then surely it's Nvidia's fault for allowing that to happen?
    I mean, if they are using standard APIs and there isn't any protection against overheating in the card firmware or drivers then that is surely Nvidia's problem.

    Fixing to a maximum fps seems like it would reduce the pressure on the GPU to work around the issue but shouldn't be required.

    • by AmiMoJo ( 196126 )

      Reminds me of when I think it read Dell claimed that VLC could damage speakers on its laptops. Installing VLC voided the warranty on the speakers if you forgot to delete it before RMAing the machine.

    • then that is surely Nvidia's problem.

      Only if it's an NVIDIA FE card. OEM manufacturers of chips leave the implementation entirely up to the vendor. EVGA is quite well known for binning chips and producing some cards with incredibly high factory overclocks and significantly higher current powersupplies tweaking the chips to the last inch of their life, which of course also makes them benchmark higher than NVIDIA's own products.

    • by Dutch Gun ( 899105 ) on Saturday July 24, 2021 @12:02PM (#61615839)

      Agreed. Professional videogame developer here, although admittedly not a graphics specialist. Videogame developers literally have zero direct access through hardware, or even access to drivers. We simply make use of common APIs like DirectX.

      But all this indicates is that there's some thermal-related flaw in Nvidia's RTX 3090 cards that the game just happens to be exposing, NOT that there's anything necessarily bad the game is doing. Really, the only thing a game can do in this case is detect this card and throttle down the rendering / shader details to try to avoid working the card too much, or just cap framerate overall (which it sounds like what Amazon is doing).

      It's sort of ridiculous for the headlines and article to insinuate that this is Amazon's fault in any way. Any other number of games are likely doing the same thing, but it would be hard to detect a correlation, unlike with a new MMO. Gamers routinely come to all sorts of ridiculous conclusions about how the DirectX calls your game is making must somehow be unique from all the other DirectX calls other games are making, and therefore it's your game's fault that their hardware caught fire.

      • by labnet ( 457441 )

        Awwww
        And I was picturing bezos with a stupid grin slowly stroking his bald cat.

      • by djinn6 ( 1868030 )

        Really, the only thing a game can do in this case is detect this card and throttle down the rendering / shader details to try to avoid working the card too much, or just cap framerate overall (which it sounds like what Amazon is doing).

        So why don't games cap the frame rate by default? It should be possible to detect the screen refresh rate, then cap it to that number. There's no reason to render 1 million fps when your screen can only display 60 of them. It's just wasting power.

        • It generally is a good idea for a game to offer the option to cap frame rate to the refresh rate, and I think it should be enabled by default. In fact, modern Nvidia drivers have this capability themselves, without relying on the game to enforce this. Running uncapped can be useful for benchmarking, or when tuning graphics options, so it's often optional.

          My point, though, is that it's a bit misleading to blame this obvious hardware fault on a game which simply works the GPU too hard to the point of (proba

  • by loonycyborg ( 1262242 ) on Saturday July 24, 2021 @09:28AM (#61615401)
    If a game found a way to fry a videocard unintentionally then it's only matter of time before some malware replicates the same issue on purpose.
  • Misleading... (Score:4, Informative)

    by msauve ( 701917 ) on Saturday July 24, 2021 @09:29AM (#61615407)
    The headline should be "RTX3090s getting bricked when used at full capacity." Can't blame the game.
  • by timeOday ( 582209 ) on Saturday July 24, 2021 @09:33AM (#61615415)

    10 PRINT "NVIDIA SUCKS"
    20 GOTO 10

    Behold the evil genius of not self-throttling!

  • Wouldn't these cards be under warranty?

    I suppose getting a replacement would take some time if it happened at all. I've had two cards replaced under warranty, PNY, the psychedelic checkboard screen of death (Only in the winter when the card was cooler).

  • And so as a KNOWLEDGEABLE beta user you would NOT leave the framerate set at uncapped on a high power/throughput video adapter of course, would you?

  • by cpct0 ( 558171 ) <slashdot.micheldonais@com> on Saturday July 24, 2021 @09:55AM (#61615473) Homepage Journal

    This is probably due to a very specific combinaison of calls that taxes one part (but not all) of the 3090, making it overheat while the actual thermal sensor is far from that taxed piece, before the cooling has time to kick. So it has time to fry. Once the game is loaded and gone through its paces, it's probably fine, as the entire card's cooling gets ramped up.

    The said part is the game vendor probably didn't have any 3090 EVGA at hand, so they probably had to purchase 4-5 of them at scalper costs to try out the game, bork 1-2 of them, and fix the issue, letting the game soak up on that 3090 for some moment. Yay for 10-15k$ worth of scalpers money.

  • by RogueWarrior65 ( 678876 ) on Saturday July 24, 2021 @10:03AM (#61615495)

    Amazon is trying to corner the bitcoin mining market so just like the Malaysian police, they're crushing the competition.

    • I'm actually kinda surprised some video game company hasn't tried this yet; only mining when the GPU isn't heavily utilized.

  • The hardware must be pretty shitty if software can destroy it. It never happens with cpus. Ooops ran prime95 for too long and now it won't boot, said no one ever.

    • It never happens with cpus.

      Errr yeah it does, and a couple of motherboard vendors have been called out in the past for setting up CPUs to boost way beyond spec causing hardware damage. And yes this is entirely up to motherboard vendors, Intel simply provide the values to use to motherboard vendors.

  • that it can be "bricked" by software that runs on it, then it's not the software's fault. This should be obvious to even the dumbest computer users in the world. Somehow, to "journalists" it's not obvious, hence they write sensationalistic, misleading articles blaming some software while not even asking a single question about why the graphics card can be bricked by running a game.
  • by account_deleted ( 4530225 ) on Saturday July 24, 2021 @10:49AM (#61615659)
    Comment removed based on user account deletion
  • by Anonymous Coward
    EVGA has already started to replace all their cards that fail. Of course, depending on how many that ends up being, one can easily imagine the wait times to receive the replacement will start going up.
  • Nvidia's shit engineering bricked these cards. Not a game from Amazon.

  • If this were specially crafted code to take advantage of an obscure fault in the care, I would blame the software, but the card should have its own protection against normal usage.

    Surely it's not reasonable to expect game designers to know the failure limits of every type of hardware (including not yet released hardware) that might run their games.
  • How about software in mask rom that the user can fall back on in case the flash based firmware gets borked?

    I wouldn't expect this in a cheap crap no name graphics card, but the high end ones should have this as a standard feature.

    • How about software in mask rom that the user can fall back on in case the flash based firmware gets borked?

      I wouldn't expect this in a cheap crap no name graphics card, but the high end ones should have this as a standard feature.

      The ignorant journalist is misusing the word "bricked" in this case. Amazon's MMO is not doing anything to the card's firmware. It's causing a part of the card to overheat to the point of failure. The magic smoke is being let out, rather than bricking the cards. And yes, nVidia should design their chips better so that doesn't happen. Thermal throttling has been an important feature in integrated circuits for 16 years since Intel introduced it into the Pentium IV. You'd think nVidia would have mastered

    • by Mal-2 ( 675116 )

      An EPROM (the kind you erase with UV light, not an EEPROM) might be a suitable choice. You can re-program it, but it requires taking a label off a chip and leaving it under a lamp, so it's not susceptible to editing after the fact. However, they could still be kept updated at the factory until either the label goes on the chip, or some other piece of the build blocks the light instead.

      I remember motherboards having dual BIOS, one you can flash and one you can't. This would be the GPU equivalent.

      • Sadly, a lot of vintage computer hardware does not work anymore because the EPROM used for the firmware has faded overtime. This is especially true for hardware for the 1970s and 80s.

        Of course, even back then companies did not expect people to be using that hardware 40 years later, and it's extremely true for today's 'fast tech'.

        Like how table top electronic games will never use a real VFD again (a real shame, those were awesome. The LCD facimilies used in today's 'classic' re-releases really don't

  • by Mal-2 ( 675116 ) on Saturday July 24, 2021 @04:09PM (#61616529) Homepage Journal

    According to JayzTwoCents, EVGA is honoring all RMAs and even cross-shipping the replacement rather than waiting for the broken one to arrive. While this is definitely the best way to save face in this situation, I have to wonder what's going to happen when their stock runs out, and further what they do when the replacement cards start getting RMA'd too.

  • by ELCouz ( 1338259 ) on Saturday July 24, 2021 @04:21PM (#61616567)
    Red light is due to fuse on the PCIe power (75w pcie not the 8 pins ones) being blown. Some games can trigger this issue with FTW3 models (3080 & 3090). For some reason it makes the GPU draw over spec the power on the PCIe slot. EVGA still refurbished them just by changing the surface mount fuse no PCB correction being done now.
  • Could somebody please explain to this dummy how can any user-mode software be held responsible for a hardware failure? I don't like Amazon at all, but IMHO EVGA, nVIDIA, and whoever else sells this faulty hardware should prepare to pay fully for these damages AND recall what still works, or expect a class-action lawsuit. BTW, what happens, if some malware written for lulz, performs the same operations and bricks the GPUs?

  • Is the video-game just window-dressing on either a mining operation or a background compute solution?

HELP!!!! I'm being held prisoner in /usr/games/lib!

Working...