Hello everyone! I detected that my GPU was hitting the thermal limit when playing COD, the game crashed and in GPUZ, the hotspot was 105 degrees, the memory was 86 degrees, and the GPU overall temp was 88 degrees on load. After some research, I ended up replacing the thermal pads (https://kriticalpads.com/msi-gaming-x-trio), followed the instructions, and watched some videos on youtube (See pictures). The problem was not solved; the GPU keeps hitting 87 degrees and the hot spot 105. Thou the memory went down 15 degrees. Doing some benchmarks, I noticed the voltage does not go up even on load (Heaven Benchmark). The fans are running at 100%, which makes no difference, I tried reapplying thermal paste, but that did not work either. I need help!
Well, can't argue that this card is a piece of.. art when it comes to cooling and overclocking. Can't help but wonder did you buy that card off of some mining rig? Because I bought mine brand new 1 week after it released and I've been kicking it in the stomach ever since but only with natural loads, not mining ones - hasn't failed me like this even once. Warzone 2 after hours of playing makes the card do 81c at gpu, 89c at hotspot, 94c at memory, and its been like that since literally day one, and it was always overclocked. The fact that you've already went ahead and replaced vram thermal pads makes me believe you got the card for cheap and now you're feeling the after effect. RMA it if you bought it new, if not - return it to the guy that sold it to you if it ain't too late. Here's a gpuz pic of how mine reacts to an hour of warzone.
Got it brand new from Best Buy end 2021. I think I was supposed to send it to MSI instead of replacing the thermal pads. Now I won’t be able to do so cause warranty it’s void since I opened it.
Your first instinct is to 'do some research' and open up a brand new card for which you paid probably near 2000$? I'm sorry but the plot holes here are off the charts. It's kind of hard to believe you'd open up the card if you paid full price for it. You literally have to poke a hole through the warranty sticker to open the card because its on one of the screws on the gpu core. I'd honestly love to play along but I've seen what mining can do to chips and I just cannot believe this is your own card. You don't act like someone who paid an arm and a leg for a gpu. And by your own words you bought it in the worst moment of gpu pricing history.You are skipping 99% of the 'leading up to' moment of the story. If your card was fine one day and hitting 90c and 100% fans the next day, as an owner of this specific model you would know straight away that something is off because this card at its default fan settings make it end up at 60% fan, if i turn it up manually to 100% the difference is literally night and day. You cannot NOT notice such a thing. If so, when did it happen? What changed around that time? Did you screw with your PSU or something? If you have the balls to open such a gpu then you should be monitoring it 24/7 when playing stuff. So you must know the time at which this discrepency happened, think about what changed around that time. These are all brainstorm pointers for you, if the card really is yours, Whichever the case is here, you'll have learned at least one lesson out of your situation. If you really did buy the gpu brand new - never open it before warranty runs out, especially if its as expensive as this one. If you bought it second hand (it was probably used for mining) - never pay before you murder that card with 100% load on the spot.
i can say he really brought the card new, i detected the issues when he was stream, the card was hitting 86 degrees all the time on load, and the fps werent that high as a 3090, so we digged in, did some test and benchmarks... so it looked like a themal issue, we saw most cards arent even well padded well, even a normal card can get up to -10C across the board, so instead of sending it and probably wait 20-30 day... we though it was an easy repair as the card doesnt have any other syntoms. Also he didnt used to monitor the card, we download HWinfo, Gpu-Z to do the test, everything except the termal, we recognize it was a mistake and that something else is going on as the temp didnt move 1C which is imposible, the only thing that is correct are the mem which came down from 90-100C to 70-80C. We know the pad and termal repaste is fine because Idle temperature was around 50Cbefore and now its around 38C, so the problem is underload. We think it was always like that because the temps and fps on warzone were always like that since he got it. Since the thermal repair was an failure, i try another thing. i turn up MSI afterburner, and locked the power limit to 60%... Now the game is giving the same fps but the temps are -10C... also the voltage went from 0.7v to 0.9v which is funny my 3070 underload uses like 1.15v, so the card is really not boosting because the thermals. also watch this video even if it was a mining card it shouldnt act that way. There is the complete history, any other question? we already filled for msi sopport ticket, but as this is a well known OG site we also shared the case here, maybe with were missing something. Also in the US is ILEGAL by the federal laws to void a warranty, simply becuase "it recived an unathorized repair" it violates Magnuson-Moss Warranty Act.
Well such a friend you are. Instead of RMA'ing it straight away you decide to void the warranty and poke around, possibly ruining the only chance of getting it replaced whilst accomplishing nothing. Even if the re-seating of the cooler did fix it, it would still be a stupid move because if that gpu drops dead tomorrow - you know whats next. In terms of the issue itself, I've seen a lot of weird gpu behavior over the years but I do remember one that was specifically odd and similar to this one. A 1080 Ti a friend had running on a 550W PSU. The core was overheating and the card was drawing like 80% of its power while the TDP limit was set at 110%. I'm not sure why this was the case but the moment we swapped the PSU with a 650W seasonic one - card started drawing full TDP and the temps/clocks normalized. There was nothing( at least to our knowledge ) wrong with the 550W psu that was in there, and there was plenty of watts left since the cpu was some 65w i5 - but for whatever reason it was making the core overheat. If you can just swap out your 3070 with his and see if it acts any differently. If it doesn't then.. someone more knowledged here has to step in because I really can't wrap my head around what might be going on. Bad contact with the cooler is literally the only thing on my mind.. try liquid metal since you've already voided warranty. I mentioned the mining stuff because I've bought mining cards for friends, 1080s - 2080s and even after reseating all of their cores' temps were so bad that the gpu's turbo clock was like 30% lower than usual, but given the price they were still worth it - mind you all of them were poorly taken care of. Linus' vid won't show every scenario possible.
What case do you have and are you getting enough airflow through it? The 3080ti will slowly cook itself and the rest of the PC if the case doesn't have enough airflow.
that was my initial though, but the case is more than fine, and the airflow is correct all fan are well distributed.. full tower case can remember the name thou.
as i said, they would have to demostrate that we broke it bro, all we did was standard maintenance, repaste and changing the pads. something else is going on, is ilegal to void a warranty just because of a sticker, lot of company are in the eye of the Federal trade commision for this. you american dont even know your own rights, that why this company keep fckn you guys up. ill check the power supply cables and try the card on another pc.
If you reduced idle temperatures by re-pasting but it still overheats under load, it must be either the GPU fans or case fans are not fast enough or the GPU vcore is too high. The only other thing I can think of is that the cooler is defective somehow. eg faulty heatpipes, not making enough contact with the GPU.
How many case fans does it have and are they 12/14cm? And what CPU cooler does it have? My bet is still on the case. The 3080ti can easily run HOT and kicks out a lot of heat to the case.
like 8 fans in total i dont know the exact case as it isnt mine... but everything seems right. it one of those case which have isntead of 3 front fans it have 3 backplate rear fans for intake. he can get 2 C diference with the case open...
