No ty voole....
*****
Did the New World Beta brick your gpu? : newworldgamehttps://www.reddit.com/r/newworldgame/comments/oobi56/did_the_new_world_beta_brick_your_gpu/This is why it is happening
So, NVIDIA partners such as ASUS, EVGA and MSI, vary wildly in design and target audience on each architecture, on Turing (2080 Ti) the EVGA FTW3 was arguably the best card out there, while the ASUS Strix OC was the worst, this has now flipped on Ampere (3080 and 3090) to ASUS Strix OC being the best while EVGA FTW3 being the worst, you absolutely cannot trust these partners for any reason, especially be brand loyal, all of them are out to make money on you and they will cheap out when they can, as well as false advertise and change target audience on a whim, thus, each card you buy has to be reviewed individually. This is obviously not something most people do (do extensive research before buying a card).
Example; ASUS Strix OC technically has the second best designed PCB and components on Turing (2080 Ti), equipped it with very strong power delivery (VRM) that can easily handle a power limit of above 600W, but at the same they limited the card to only two 8-pin power connectors which limits it to 375W max to stay within ATX spec, but they didn't even let the card use the full 375W, instead they capped it at just 325W which literally makes no sense on any level (extended triple fan triple slot cooler), since the Founders Edition (cheapest VRM you can get and a dual fan two slot cooler) had a power limit of 320W, so you paid approximately $300 premium to get 5 watts of power limit (performance), so overclocking a Strix and a Founders Edition you ended up at the same performance, give or take a few % because of the lower temperature on the Strix.Meanwhile, EVGA on their 2080 Ti FTW3, also a strong VRM that was capable of above 600W, still used two 8-pin connectors just like the Strix, but they (EVGA) let the us use all of the power they could give us while staying within spec, 373W out of 375W possible. Thus the FTW3 performed far better both stock and overclocked, 48W higher power limit, and the card cost $50 less. And what did the market look like? Strix OC outsold every other card on the market here in Europe, people don't/didn't care the least about if the card was any good, they liked how it looked visually, the cooler and backplate, as well as found ASUS reputable for some reason (brand loyalty/fanboy), so the summary here is that you literally need to do some research before you blindly make spend over a thousand dollars on a graphics card.
Okay, so what changed on Ampere? A lot.. ASUS didn't cripple the Strix OC, right away we got a power limit of 480W, triple 8-pin card, which allows for a maximum power limit of 525W to stay within spec, so not all the way but close enough, most games don't use above 400, even when increasing the voltage, software absolutely affects power draw.And for EVGA on the FTW3, they offered just 450W, so 75W shy from the maximum spec. Now, here is where the issue occurs, components are great, the cooler is great, the price is fantastic.. but the power delivery DESIGN is awful, bordering on insane.
(Optional): Read this before you continue if you really want to learn
https://www.overclock.net/threads/how-nvidia-neutered-the-rtx-2080-ti-with-intrusive-power-limits.1753844/So, most of you surely don't know, but both the FTW3 and Strix OC are LN2 cards, they are not meant to be air cooled, that was literally not what they were designed for, these cards can without struggle pull 1000W for long periods of time. But it's a marketing strategy, make a large (extended PCB) visually appealing card, throw a bunch of overkill components on it and a massive cooler, and a huge price tag, and people for some reason flock to it even if they're never going to use any of the features (the strong power delivery & LN2 BIOS). Here is a big truth which I'm definitely sure none of you know, on Turing, the PCB and power delivery that NVIDIA designed, is so overkill, that there is no reason to buy anything other than the cheapest card you could find, the absolute most basic model NVIDIA offered, was $999 MSRP and featured 13+3 VRM capable of running 600W on LN2 BIOS, it was so good that you could run it with an AIO cooler at 400W, with no airflow or heatsink on any part of the card except for the GPU, zero airflow over the memory and VRM components in a 30C ambient room, VRM components reaching 100C, it was/is completely fine, that tells you, truly, how overkill NVIDIA made the 2080 Ti Reference PCB (power delivery), thus, if you or anyone you know paid for a model above the most basic model available, you/they wasted money, and sure you might've gotten a lower noise level on those massive coolers, but was it worth several hundred dollars? Also, these custom PCBs designed by ASUS, EVGA and MSI, are not nearly as validated as what NVIDIA offer, the odds of anything going wrong or a design flaw, is far greater on these models, personally I had not one, not two but three EVGA Classified cards a few years ago that all died because of flawed design, warranty had ended 3 months earlier, so was left empty handed. So what I'm saying is, if you stick to the "tried and true" NVIDIA design, risk of something going wrong is greatly reduced (This is exactly what happened to the Ampere FTW3 design, they messed up bigtime).
Finally, now you should have a grasp on why you should or should not trust these expensive custom designs by ASUS and EVGA.
This is what EVGA messed up on the FTW3 on Ampere (3080 and 3090), the short answer is that they cheaped out on components, they opted to pick literally the worst controller for the PWM, which is analogue as opposed to digital, what this means is that the controller being analogue, cannot change, it's programmed for one specific power value, which is 450W as I mentioned earlier, this means that you cannot increase the power limit because the controller is locked to 450W, so when EVGA released their updated 500W BIOS to compete with the 480W BIOS that ASUS shipped with the Strix OC, that 500W BIOS update didn't actually do anything, and the severely incompetent PR people at EVGA dared to call it "XOC" which stands for Xtreme Overclocking, which is a term specifically used in association with LN2 overclocking (competitive), which this BIOS is obviously not, on LN2 you need at least 1000W, so they missed the mark by quite a bit.. this is nothing but incompetence and false marketing.
More about the controller, so, this analogue controller can be referred to as a dumb controller, it's there only to do the very basic functions, while a digital controller can be referred to as a smart controller, advanced features as well as an unlocked power limit (can be adjusted), meaning you can essentially flash a 2000W BIOS onto any card with a digital controller and you will have effectively removed the power limit, the card will not throttle for any reason, even temperature as that is typically disabled.
Okay so, what does this 450 number have to do with cards dying? That is not the first mistake EVGA did on the FTW3, they also added fuses to all the shunt resistors,
Shunts are, as the name suggests, a resistor that is installed in a circuit in sequence with the actual load and in parallel with the measurement device. This resistor has a fixed value, like 5 mOhm in this case, and serves as a reference for the GPU to calculate the current and power. What the GPU also knows is the voltage present before the resistor, e.g. 12 V at the PCI-Express connectors or in this case the Nvidia 12-pin Microfit connector. This voltage is simply measured before the shunt resistor.The resistor now causes the applied voltage to drop, linearly with the current flowing in the circuit. If the actual load consumes more current, the voltage at the shunt also drops more. Behind the shunt the voltage is measured again and from the before-after results a delta, the voltage drop. Since this is linear and proportional to the current of the parallel load, it is now possible to directly infer its power consumption.
Character Limit, continue to read in Part 2 of 2:
https://www.reddit.com/r/newworldgame/comments/oobi56/did_the_new_world_beta_brick_your_gpu/h601e1y/Let's say, these shunt resistors are connected to the dumb controller, each shunt resistor is allowed X Watt to go through it, so when EVGA shipped the 3090 FTW3 with the 450W BIOS, the controller itself and shunts were all locked to 450W, and there would be nothing you could do about it, so the FTW3 has three 8-pin connectors, each allowing for a maximum of 150W (to stay within spec), and the PCI-E connector on the motherboard can supply 75W, this gives us a total maximum of 525W, but EVGA locked it to 450, so let's say 66W for the PCIE, and 128W per 8-pin connector, 125+125+125+75 = 450W, now, if you flashed the BIOS to the new EVGA 500W one, they changed the values to ~142W out of 150W possible, while the 75W was already capped so they couldn't change that, this results in 142+142+142+75 = ~500W, sounds good? No, because the dumb (analogue) controller has a fixed ratio, if the PCI-E is set to 75W, 100%, and each 8-pin connector is set to 125W 100%, increasing the 100% 125W to 142W would make that % go up to 114%, this is not possible, it cannot go over 100% to stay synced with 75W PCI-E, so even with the updated BIOS released by EVGA themselves, telling you that you now can use 500W, each 8-pin will still only allow 125W and the PCI-E 75W = 450W.
Okay, so this is obviously very very bad from an overclockers perspective, they literally locked the card down hard, this card cannot be used for LN2 anymore, all of those expensive VRM components you paid money for are useless, the card will never go above 450W because of the dumb controller. But that didn't stop people from trying, so they modified the shunt resistors, either by replacing or altering them, to make them read differently, this effectively made the same change as the BIOS update, if you tricked one of the 8-pin shunt resistors to read 50W, so it could pull much more as it wouldn't be detected as reaching the limit, that did not work, proved to be just as ineffective as a BIOS update, so what people finally figured out, was that you had to modify all four shunt resistors, the PCI-E and all three of the 8-pin connectors, when all of the 4 were changed from 100% to read 75% each, then and only then did the power increase, great? No.. as soon as people finally unlocked the power from this dumb controller by tricking it in all four locations, EVGA had added a sea of fuses, and these fuses were rated just above the stock 450W, so when you started pulling 600W, those fuses popped instantly and bricked the card.. This is what is happening in "New World", for some reason, that even I can't explain, something about the software is causing these fuses to pop and bricking the card, nothing is actually damaged, as said, these cards can easily handle 100C on the VRM as well as have 1000W pushed through the card, without damaging it. Games (software) vary wildly in power consumption, I can run GTA V in 5K@60 and my 2080 Ti (with unlocked power limit) will only consume around 325W, while if I start up Metro Exodus, the power consumption will peak above 500W, that's a massive difference in both power and heat. So again, for some reason, the game New World is causing something strange to happen to the power consumption, the card is still very much locked to 450W on the 3090 FTW3, it cannot go above, as well as temps are clearly fine on that massive cooler, zero blame can be put on the developers of the game, sure they can have spaghetti code but it should never cause damage to a graphics card. The bottom line is that EVGA clearly made several errors with their design on Ampere (3080 & 3090), and we know that the fuses are the only explanation that sets this card apart from any other card, that can cause cards to suddenly die (again, they're not actually damaged, just launched into safe mode to protect further damage).
Again, without actually seeing all the data, such as power consumption, temperatures, GPU and Memory utilization, it's very difficult to even guess what is causing the fuses to pop.
There is one piece of good news that I need to end with, EVGA isn't completely incompetent thankfully, they actually changed the analogue controller to a digital controller some months after release, so every new card sold today is using this new controller, and will handle the 500W RBAR BIOS just fine, fuses won't pop and you'll get all of it (500W) safely. Also note that I never mentioned 3080 Ti above, only 3080 and 3090, because 3080 Ti launched much later and thus only have this new digital controller, so no Ti (FTW3) models exist with the old analogue controller.
Also, these fuses can pop even on the 500W BIOS on the old cards (not just in New World), it is obviously covered under warranty so you will get it replaced with a new card with the digital controller, it's just annoying but don't be worried that you'd lose the card, it will be replaced.
So tip for the future, to everyone, DO NOT trust these companies to sell you quality, do NOT be brand loyal, these companies change personnel and designs all the time, as well as prices (product segmentation).
TL;DR: There is no TL;DR, read it.