Same here my dude, now i'm in mortal fear that my beloved 6800XT Red Devil will kick the bucket and AMD only has the 7900XTX as a top tier gpu; also i don't want to pay nGreedia for a 4090 :(( I'm scared shitless rn :((
@@royvice666 The 7900XTX's seem to be fine so far, just my own personal opinion/ experience, Judging what what I've seen on this cannel a lone you're more likely to encounter a dodge 4090 than an XTX and I suspect it's due to the rush Nvidia put on pumping out 40 series cards after their 20 and 30 series launch debacles.
I have a 6950xt and a 6800xt both are faster than Nvidia cards in that timeline due to faster rasterization so no worries, dude just identified himself as an idiot with no knowledge that Nvidia cards were only faster if you used ray tracing and we all know no true gamer uses ray tracing, however it is very common for repair guys to badmouth electronics that they cannot fix or understand
A few tips for all the 6800/6800 XT/6900 XT owners scared by this video. 1. Even if your VRAM is stable at the max 2150MHz, do not run the VRAM at 2150MHz. Stay in the 2100-2120MHz range. It's much easier on the IMC (which is fragile), and the performance difference is negligible. 2. If you're seeing more than a 30C spread between edge and TJ temp, stop gaming and repaste the card. If the TIM coverage is poor in the wrong spot, your core may not be long for this world. 3. UNDER VOLT. UNDER VOLT. Keep within the 1000-1050mV range on the core voltage. There is NO reason to ever run the stock 1150mV. 4. If your TJ is reaching 90C, don't surpass 280-300W for long periods. Don't let TJ surpass 100C at any wattage or any length of time, as whatever small performance gain you're getting at 100C TJ isn't worth it. Do all of this and your 6800/6900 series card will continue making the value proposition of RTX 4060/4070 cards look very poor for years to come.
@@devilzuser0050 50MHz doesn't seem like a lot, but if you encounter a dead 6000 series card, chances are good that the owner will tell you they always ran the VRAM at 2150MHz. That 50MHz makes all the difference, which I agree sounds absurd. But there is a reason AMD hard capped the VRAM at a hugely conservative 2150MHz when the particular IC's the 6800/6900 series used were all easily good for 2250MHz+. The closer you get to 2150MHz on the original 6000 series parts, the greater the risk that the IMC fails. The Radeon 6x50 XT parts were a revision made to correct IMC fragility and they could better leverage the GDDR6 ICs being used without risking an IMC failure.
@@fdgdfgdfgdfg3811 No, it's not. If the core kills itself it's not consumer's fault, and if the stock card has hotspot issue from factory is also not the consumer's fault.
Stunning that most of the comments seem to completely miss the point that he's criticising AMD's GPU hardware reliability and quality of this GPU's core while you guys are all talking about the 6800XT's rasterisation performance.
Stunning that he just said that 6000 series GPUs are bad without actually saying WHY they're bad, or giving any evidence other than just showing us a single example of a dead GPU. He may be correct, but he has to at least say WHY they're bad, or give some more evidence. Are they failing at a high rate? He needs to be able to give some estimated figures on failure rates, or explain more specifically what's bad about them.
I have a bunch of 6800 XT, 6900 XT and 6950 XT cards running here (XTX and Sapphire) not a single one ever got over 90 °C under full load. The last time I got a problem with any AMD card was, let me check my records, a XTX RX 5700 XT Thicc II that was DoA 5 years ago.
@@DJDocsVideos same here, I'm still trying to figure out what kind of case/environment these folks are running their cards in that allow them to get that hot.
Others go for max profit and that requires only attempting fixes that are sure to succeed. It's more more profitable to report "not fixable" for borderline cases and still bill for diagnosis fee. Of course, you still always complete all the easy fixes to maximize profits.
@@MikkoRantalainen the issue is the person am talking about takes fixing fees regardless of him working on it or not. (just inspecting) and still charges full price.
@@adamtajhassam9188 Lol, wtf are you talking about? reliable card? he just said both have their flaws, AMD 6000 seires bad board, 7000 series have good cores but bad boards, NVIDIA series 900 and 2000 have crap cores, however other series have bad boards which is where they fail... what are you talking about? reliable card... want a reliable card? go back to the years 2000 or just don't overclock it.
@@adamtajhassam9188 you didn't even watch/hear the commentary in the video, lol. he just said nvidia cards are bad, only cores are good. opposed to amd where cores are bad.
@@dumbdoggo hard agree. guarantee owner of this card modded it or was blasting frequency or voltage more than he should have, probably by a disgusting amount.
Backstory on the card is: Purchased used from eBay in Dec 2022, from United Arab Emirates seller, most certainly it was mined or refurbished card (one memory chip has different serial number which means it was repaired before), card became no detect in July 2024. Card ran stock, light gaming, not demanding games, Seasonic Platinum power supply, case with custom, good air management, core temp in normal range, hotspot could have been better but nothing extreme, on higher side of average.
Tony your absolutely 100 percent honest in your reviews. You're a tough sell though ill say that. Thats understandable though. Your opinion on most things is that they are garbage... if they are in fact garbage. You hold these manufacturers to a level that expected of them. Its their fault they sell crap.
They have been doing it for decades, and they know that 90% of people who buy in this tier (upper tier) of GPU are gonna upgrade before 36 months, right? So it is what it is. Doesn't change the fact that the 6800XT was an INCREDIBLE card ... for GAMERS. Who play games. If theres a fault before the upgrade occurs theres RMA, theres returns policies if you buy from any decent manufacturer like Sapphire etc. Sometimes 5 years warranty (on most Nitro cards). XFX lost their prestige years ago IMO. Used to be right up there, but certainly not these days. They are a budget option for a reason! Whenever I buy any AMD/Radeon card, I'm 100% getting a Sapphire. Been that rule since the Polaris days. Never had a problem either!
I’ve had a mixture of Nvidia, AMD and even Voodoo boards going all the way back to 1996. I have never had a single GPU failure. Zero. None. Out of at least 30 GPUs over a 28+ year period. All I’m saying is that being in the repair business doesn’t give you a good overview of reliability. *All* you see are failures. What you don’t see are the 99.99% that *don’t* fail. In my field (medicine), this is called Selection Bias.
I guess I've been really unlucky then. Out of the five cards I've had, two have failed, one in less than a month. Both cards were AMD, while my Nvidia cards have never failed. I love AMD CPUs but their GPUs have left a really bad taste in my mouth...
I'm about 8 years behind you and same, except for an 6970 card I screwed up some capacitors on and couldn't fix and a 7970 that was run daily with literally buried sliders (except VRAM/just before ECC kicked in), 40% voltage and overclock lmao, on a stock blower and even did some early bitcoin mining tests. It was ridiculous that card, the goat of the original AMD days.
Yeah. This is how every industry tracks reliability. What you have in an anectode. What he has is a massive dataset of many GPUs coming in for repair. If he's seeing more 6000 series than say 3000 series (nvidia) then that data is true. Sure, not all 6000 series AMD gpu fails, maybe 2% do. But remember, there are roughly 3 times more Nvidia GPUs in the wild than AMD GPUs so if AMD are showing up more than Nvidia at repair shops, then its way worse than we think
Same here, used dozens of cards, both green and red ones over a 20 years, built several PCs for friends and family too, and my only dead GPU ever was some crappy one (I think it was 8600gt) with passive cooling I've been using intensively for 3+ years😂
Been running this exact XFX 6800 XT Merc 319 card stock settings since Jan 2021 and never had an issue, and still don't. Never undervolted, never overclocked. Temps have always been fine, still are. Typically gets to 70c edge temp and around 85c hotspot temp in heavy loads. Fingers crossed.
Don`t hop on the "OC-Train" ! DON`T oc EVER !! It`s not worth it ! Undervolt by a bit, repaste the GPU every 6 month .... and for most of the time you`re golden !
Na run that 6700.. not all have the crap vram. If your oc with a hack yeah you can kill your core pretty easy. If you take care of your gpu. It will do you good.
If it works fine for you, why bother? Even if he said "these are generally bad", it doesn't mean that your core is on a timer and will inevitably explode.
All you need to do is gently undervolt and underclock them. It's the same on many Nvidia cards: RTX 20XX pretty much requires it for any GPU with a bit of use on it, and RTX 30XX has huge power spikes if you don't do it. The power spikes will crash your computer if your PSU can't handle them, and even if it can you will kill board components over time. If you're upset you bought a lemon, know this: so did everyone else. We need to demand more from AMD/nVidia
@@FlakAttack0 100 percent.. Don’t game with it at its max fps, always on the edge of 100c and it will last you years. Normally I do test it for its peak then turn it down, under volt a few clicks until it’s steady 60c and hot spot 80c. 5xxx 6xxx and now have a hand full of 7xxxs none have failed me.! My only issue was with the thermal paste getting hard as a rock on 6xxx and 7xxx series. We do replace the paste and pads every year or so.
@@absolute___zerodidn't need to do that GTX series. Only RTX series are failing like this theyre nor even that old things like a gtx 1080 to are lik 7 years while these are two or three years.
As somebody who has been on the design and manufacturing side it's not on purpose. It's more a "happy accident". Designing boards to prevent the failures this channel shows spikes costs like you wouldn't believe. But it is easy enough to pay maybe 50% more to just overengineer the board so it never overheats. Would you pay an additional $150 for an overengineered board?
@@julianmorgan79 yes always. AMD Intel or Nvidia whatever the piece of hardware it is , I always optimize doing my research. Optimize ( performance per watt on CPU) For example undervolt and PBO to negative. For GPU ( VF curve) I experiment with my silicon, at which lowest power it can provide me with stable uncompromised performance. For my 6800XT i Use 950 to 980 mV at 2250Mhz . My stock settings are 1150mv at 2400mhz Gives to 20/30° lower temps with almost half (if not optimed 250W to 300w consumption) 150w to 200w max consumption.
So, I have a 6750XT, and I did notice something about it back when I first got it, that makes me wonder if I would be dealing with the same situation right now if I didn't go and fix the usually atrociously applied thermal paste on it. It ran hot, for no good reason. Even when the fans were turned up to help combat the heat output, it still ran kind of hotter than I would expect it to. Not sure why that is, or if it has anything to do with this 6800XT failing. But ever since I put new paste on, my way instead of the 'proper' way that supposedly the manfacturer also uses... now it runs fine. No overwhelming heat. No sudden crashes to desktop during gaming. Nothing. (Crosses fingers hoping not to jinx it.) Now, I'm no repair tech like you are, but I am pretty sure the core getting hotter than needed on a ongoing basis probably isn't too good for its electrical connectivity, especially when all these comapanies are choosing to use a lower tier solder in the name of the environment and probably savings in their profit margin being their real motive. A solder that is known to melt easier at lower temperatures than regular solder. But surely it can't be getting hot enough for that to happen, right? I mean, who lets their GPU run full tilt in conditions bad enough to let it get above even 80c? I'll tell ya who. A dumb monster who doesn't deserve to have a computer. Or maybe a crypto miner... which doesn't really change the outcome consequence.
Laptops tend to struggle a lot with managing heat due to size constraints, so some will regularly hit 80-90c or higher. Only way to combat it is regular cleaning, replacing the stock thermal paste with phase-change pads or (very carefully) liquid metal if it doesn't come with those already, and a cooling pad. Oh and undervolt, but that's a universal thing that everyone should do regardless of platform.
@@Rayjacker Right, laptops... So, yeah they can get pretty toasty when run at full tilt, I'll give ya that. And replacing the thermal paste is once again basically a default; though I wouldn't go with the liquid metal. Mostly because of the potential for messing things up. And yes, I am aware of the methods for failsafing that, like using the acrylic clear coat for nails on the exposed circuitry to protect it from liquid metal. But let's face it, if you are having to LM your laptop to keep below 80c, it's probably not the laptop at fault completely here either. I say this, because my laptops tend to hover at maximums of 65-70 when at full tilt, and I only use normal paste. The kind provided by Noctua to be more exact, NH-1 version. No thermal pads changed on anything else, except maybe replacing them with paste where it is possible due to clearances being able to be adjusted carefully, and a little extra pressure applied where possible without breaking anything. Basically I just finish the manufacturers job so its done properly. That's it, that's all. (Proper by my standards at least.) So if a laptop I am using at least, is ever getting up into the 80c range or higher, I'm turning down the temps in the room I am in if I can. Because it's probably the room or just the temps of the day itself being too warm for whatever reason. Remember, most tests are done by folk in ambient temps, so about 21-22c. If you are in 32c temps, then don't be surprised if your laptop is experiencing a delta of 10c extra as well. At which point, the only real solution left to you, is turning down the settings. Like you said, undervolts help. I concur. They often have them set higher than necessary mostly because of people wanting the absolute most clock speeds they can get at stock. Which is nice, but has its issues. As experienced by Intel recently with the 13 and 14th gen chips specifically. But undervolting can provide a new problem. I think the term is called clock racing. Essentially, the clock speed doesn't quite have enough juice, but the higher headroom for temp threshold via lower power draw also allows it to clock higher. This can be of negligable effect on the end user, but taken too far and you start to get regressive returns on actual performance despite numbers being higher, and lower, respectively where it is normally ideal to see them higher and lower. This problem can probably be most easily seen with the X3D chips from AMD, since it's a common tactic to change the voltage values to allow for more thermal headroom. It does help, a good bit actually, but taken too far and you get that racing problem. There is one more thing. Windows power settings. In them, there is a 0-100% option for both minimum and maximum CPU performance. A lot of folk will crank these to 100%, which is fine if your thermal solution can handle it. But, if it cannot, or can only barely handle it; I suggest they try a min of 1% and max of 99%. It will reign things in, but it's not going to be liked by those chasing the highest framerates at expense of having also the highest temperatures. That difference of 1% on the max, turns off the automatic overclocking for whatever system you are using, essentially. All boost goes bye bye, essentially. I'll let this be at max during colder seasons, and such; but turn it back to 99 when in any temps above 27c roughly. With laptops I mean. Desktops... just add another fan, maybe turn them up a bit more, etc. And again, at a certain point, ambient temps are the problem more than the thermal solution. So overall, I tend to keep things on my systems in the 70s range when going 'full tilt', though I do have some restrictions in place for good measure. They don't reduce performance at all, or much at all; while letting me reduce the temps enough to justify the changes. And again, no liquid metal in use anywhere. Why aside from the potential issues I clearly know how to avoid anyways? Because the only real appreciable returns you get out of liquid metal are going to come from direct die application. For laptops, this is easier to do since they are already direct die. But desktops not so much for the CPU, even if the GPU is direct. And then on top of it all, laptops just get moved around more, which increases the chances for LM dripping somewhere it shouldn't if applied even a little bit too copiously. The phase change pads though, I have considered those before. But no one seems to be getting better results with those, than my personal method of applying paste. To keep that short and simple, I basically follow all the rules implied by the way thermal transfer happens regardless of people's opinions about best practices. Paste needs to be as thin as possible where able to be thin, because the thickness decreases transfer as it increases. But you also need to make sure all the micro grooves and fissures and such are all filled in as well as possible to increase contact of the paste and all metal surfaces of the heatsink and the IHS. And then also have to do this all in as ideal a way that can be done without using too much paste because of pump out (which can affect LM as well, hence the drippage problem.) Paste at least isn't the worst enemy with drippage, but my method reduces that as well. So i could probably get away with using LM, but I just don't wanna risk it. not without some way to make sure it's completely fail safe. Which, is technically possible, but includes a lot of painting of components with clear coat, etc. Anyways, sorry about this being so long, but yeah, I have some things to say about it all. And there they are.
@@TheKirik71 Sorry, but yeah, it's probably luck on your part. You might have gotten the one good application of paste on that run of cards. All my AMD cards are sapphire brand, aside from one Asus brand card I had a good decade and a half ago or so. It was no different in this regard. And while I am at it, it's the same situation for the AMD workstation card I have, which is technically probably a sapphire made card again (due to how the whole deal works with AMD and Sapphire), and an Asrock card for Intel's ARC series. It also needed a replacement of the stock paste. None of these were cards that were sitting on shelves for a long time either. They all needed immediate replacement, because of how terrible the application method that is generally used by almost everyone, really is. To put it simply, if you took noctua's NH-1 and NH-2 pastes, NH-2 being slightly better in results normally when compared via same application method and same ambient temps; you would get NH-1 performing as well as NH-2 using my method where NH-2 is still using the 'bad' method. And NH-2, last I used it, got even better results than normal using my method. My method is quite simple. It follows all the rules of how thermal transfer works to the letter as close as possible, without taking any lazy man methods to it at all. It's meticulous in its application, but the results are worth it, every time. But that meticulousness of it, is why manufacturers will never use it. It increases the time taken to apply the paste by multiples, because you aren't just squeezing some on and hoping the pressure works out in your favour. Because that's all the usual application method really is. A hope and a prayer. It also kind of ends up wasting some paste, because there is ultimately some loss due to the method. Some paste always remains on the applicator I use. I just saranwrap it though, and use it on the next system if it's still fresh enough. So not a huge waste, just not ideal if you want to keep it all in the tube otherwise. Results in temps dropped? Simply put, if normally reports of 2-3 or 4c is seen, with the low end being the more usual case; my method is usually getting the top end or a degree or two better. And this is all before adjusting any stock fan curves. That said, heatsink used will always affect things. But that said as well, I still see the 'best' returns from the paste in use, even with the stock intel heatsink. Granted, that heatsink isn't going to be doing that job for long, since it's going to get replaced anyways with something better regardless. Just figured I'd include it in all of this as well. Anyways, like you said, it is your first card so you can't really compare. So that's why I said all of this for you. So you can compare a bit against someone else, as well as the reasons why I am getting the results I am, from as far as I know and see things at least. Edit: I forgot to list the actual cards. AMD: 6750XT, WX 5100, 580, 290x Intel: Arc 380, and of course integrated stuff in cpus where applicable. Nvidia: Laptop gpu's. 960M, and I forget what the old 2010 laptop has, but it's 512Mb of vram. I think it's a 9000 gt series, maybe. I'll have to boot it up and look again, or find the specs online. It's not in use often anymore aside from projects I play around with where it's a good addition to said project, mostly due to the parts it has.
He’s probably not wrong about core , he does repair them so I’ll take some of his word , it can be user error though a lot of 6000 series owners overclock like crazy and I know they do so don’t say I’m wrong
a bruto overclock gains you nothing on a 6800 class card a smart undervolt however does wonders and if you got a bit luckky with the card you can outperform a stock 6900.
How can you overclock this GPU much. AMD locked the core and memory clocks in the driver. Certain 6900XT models have alot higher headroom , but not this 6800XT .
@@fleurdewin7958 Huh... My 6800XT Red Devil goes up to 2.6GHz core clock... And u know what i have seen fkers with even better samples that their cards did go up to 2.7GHz and some even near 2.8GHz, that is insane for 6800XT, as that card on 2.7GHz gets the RTX 3090 levels of performance in raster..
Not a big deal just Undervolt it using the drivers and make sure to save the profile in a easy to search carpet just in case it losses the config. So you don't have to tune everything again, just load profile and that's it.
my wife has had hers for over a year and no problems. we always babysit heat with MSI afterburner, make sure your fans are responding properly to heat ... and if you had a bad card or a factory defect, you probably would have noticed within the first 6 months.
Tony, can you do your own review on thermal past and pads for us? Yes I know there are many who have done this before but sometimes I’m not sure if they are brand fanboys or not. You just not do good fixes you are also very entertaining to watch. Thank you!.
Let us remember, AMD and Nvidia both have their dies made and packaged by TSMC sometimes on the EXACT some processes, the answer as to why certain designs fail the way they do is a lot more nuanced then hurr red man bad or green man bad. In any event, high performance, high power and high thermal load parts like graphics cards MUST be properly mechanically and thermally supported by your PC enclosure, big ass 3-4 slot GPUs without a tail support sitting in a glass fishbowl case roasting all day wont live no matter who made them. As the venerable old redhill guide stated (WRT to early MFM hard drives) "A good case can keep a cheep drive working well for years, but a poorly designed case will shorten the life of even the best drives on the market"
I had 2 different 6800 XTs, Asrock Taichi and Asus TUF. I repasted both with PTM 7950. Also repasted my 7900 XTX with PTM 7950. All work flawlessly and its been almost two years. :)
@@The_Noticer. Yes, especially with the XTX. I actually tried like 3 different pastes with the 6800xt, before i heard of ptm, and they all pumped out in like 3 weeks. PTM has good contact so the hotspot is a lot lower, by like 7c, and it simply doesnt pumpout.
@@ДмитрийКа-я6б на современных видео картах есть такой еффект, от разницы температур между цыклами работа/простой термопаста выдавливается по краям чипа и там застывает/остаётся. Лучше всего ставить pcm thermal pad на чип.
Just your 6700-6900 card under 70°C and slightly undervolt it. Better performance, less power consumption, less heat. I have a 6800 and it's the best card ever. Also love how easy it is to play with amd software.
I remember my 6800XT. It was and still is a fantastic GPU. Also bought a 6750XT for my sis' PC and undervolted it to run at 135W all the time. All that at less than HALF the price for the same performance of NSHITIA.
I do not hate you for being serious, only saying the core is bad, but WHY you consider the 6000 cores bad would be a little more informative and substantial?
i would love to watch you talk about your experience with gpus, from like gtx700 /r9 200 to modern day, what was the best/worst to work on and all about your experience with AIB , who has garbo cards/ best cards overall
I bought a brand new RX 7900 XT fom XFX not a week ago , should i be worried? how can i make it not break? (The "Brother, may i have some oats?" reference killed me lol)
Just make sure your temps are good undervolt if your worried I think xfx has a 3 year warranty use a GPU brace he said the 7900xtx has a good core bad board you have the same core as a 7900xtx/GRE so just be careful with the board
@@worldwar208 I used the brace that came with the card and for good measure I put something supporting the brace from under it. Hope it's enough, it's a big and heavy card.
That was educational. Thank you. I have a 6700XT from MSI that can't do more than 2418Mhz without getting hot (hotspot temp) to the point that it's risky. Adrenalin drivers often try to push it to 2700+ after a reboot or wake from sleep. My question is what hardware limits the card to 2418, when the chip should be able to push higher? I see some versions of the 6700XT where the upper clock limit is higher. Also, is it likely hardware or software that's trying to change the clock setting when rebooting or waking from sleep?
The 6800 xt is honestly one of the best purchases I've made, bought it a few years ago now and its still running the new games at 1440p high refresh rate. Not to mention undervolting the card makes it pull like 150/180 watts underload which is half that compared to the performance comparable 3080, in terms of repairability it might be bad but, as a gaming GPU it's been amazing to me.
6900xt here, undervolt, made her some anti-bend struct and everything should be fine in my opinion, if the card is insanely loud and hot all the time it's obvious it will eventually break itself, not mentioning faulty ones, but that's not the majority at all like with intel cpu's 😛
just because your 6000 GPU didn't die doesn't mean they aren't dog d**t. i went through THREE 6900XT. not one, not two, but THREE. first one flickered with black screens constantly. 2nd one would crash when using any game or app that consumed a lot of memory. overwatch would be fine but cyberpunk would cause the thing to crash. 3rd one kinda worked but would still crash to desktop every once in a while. ended up switching to a 3090, used the same computer. didn't do a fresh install, didn't use DDU. just plugged it in and haven't had a single issue since.
I have to say that i decided to, and bought, an MSI 4070 ti super gaming x slim due to one detail you once said about MSI boards, they put a good amount of fuses on them. 😉
@@krypz0n Yeah I did, far more likely to encounter failure in a 6000 series than a 7000 series and Nvidia is worse when it comes to board condition/ longevity. Why? What's so hilariously obvious that you think I require enlightening to exactly?
It would be nice to know about why he thinks they're bad. Without some failure rate data, or an explanation of what's wrong with the way they're designed or made, saying "they're bad" isn't very informative. The GPU failure rates could be a lot higher than other generations but still be relatively low. He gave us no detail at all.
@@syncmonism And that's perfectly fine if you want those parameters met. I on the other hand am happy with his opinion on the subject because I am sick shit of hearing content creators gloss over bad work of major brands. Plus with all the crap that went down between Intel. Gigabyte, Asus, and EK it was only a matter of time before AMD got called out on something and this is it. Their previous gen was overworked and underprepared to survive more that a single rotation of graphics cards.
They are fine stop bitching for no reason. Just get a used one with a good cooler, repaste it and undervolt it and its fine. Did the same with a 3080 and rec. A buddy to do it with a 6750xt 1+ year ago. If u don't trust me just look at the rma % on sites like mindfactory, if they where all failing left and right they all should ve a 5%+ rma rate
Yes my 6800xt red devil has bad hotspot temps like all 6800xts but what really helped kill those cards is the criminally low fan curves from the factory for the first couple years. This was due to AMD saying the cards were good to run up to 100c hotspot which they are not. This was very evident when the fan curve for my card was updated. I’ve always run a custom fan profile and monitor hwinfo64 daily. Even then I get 80-84c hotspot and 56 edge at 281watts hitting 2425mhz. Want to repaste with thermal grizzly but also don’t want to crack it open since it’s working great.
Can you or are you able to purchase cores as a replacement part within the business you do? Or is it more of a swapping of one to another as available?
@@FlakAttack0 It's not that simple as the GPU is legally recognized as a subcomponent of the PC and is replaceable via upgrade or warranty, it legally counts as being repairable as is. This is the problem when you pester governments for laws, big gaping loopholes like this are created and it's a balls and a half to make them redress the issue
Here in Dubai the local XFX retailer has been offering refurbished 6000 series cards in high volume since Q3 2022. I've always wondered how they have such high volume for the refurb cards (seriously, they never seem to be out of stock DESPITE them being an insane deal), but this video kind of explains that. Personally I've been running three refurbished cards - a 6700XT, since Q3 2022, a 6800XT, since Q1 2023, and a 6650xt, since beginning of this year. The worst issue I've had is the 6650xt suffers from horrendous coil whine. But I have heard anecdotally from a guy who buys tons of these refurb cards for flip builds, that the latest batches of refurb 6800xts have been problematic, with lots of premature deaths, high hotspots, etc. But maybe that's related more to the quality of the refurbishment going down. What's insane to me is that the XFX retailer still seems to have a greater stock of refurb 6000 series cards than the refurb 7000 series cards (7800xt + 7600) they introduced at the beginning of this year.
there are lots of positive comments about the card in this comment section, if you read. don't forget this guy sees only broken boards all day and there is really no sense of how many that is comparatively to what has been purchased. this is a popular card. my experience: wife bought one about 1y ago. she wasnt ready to spend as much as I did on her GPU, so 6800xt was her choice. no problems at all, great performance, in both heat and power, obviously the price just dunks on nvidia. i really would like a better explanation as why the "6000 gpus" are the "worst ones ever made". just keep a close eye on it (with msi afterburner) especially in the first 30 days, and if you have a problem hit up the RMA or return policy from the vendor ASAP.
Just give it a good once over, clearing any dust, replace dried pads and thermal paste and undervolt it and it will be fine. If you aren't comfortable doing this yourself try finding a reputable tech who can do it for you.
Never understood why the OEM’s took a 6 and flipped it over to reveal a 9 then changed the safe operating temperatures of their products. I may be old but it’s always been 65c is the sweet spot. This 95c-100c is the new normal is why these things incinerate themselves. BTW for Tony: I’ve been using the long q-tip type with the wood pole they work better for me, cheap and the cotton doesn’t fall apart. The kind you would see in a veterinarian or doctor’s office to clean noses and ears. Just wanted to pass the tip. Amazon has them in bulk not expensive at all.
ngl this kind of thing is what worries me about buying an old high-end card. doesn't help that the boost algorithms on modern cards gives a layer of complacency as to how high you're pushing the OC as well as the voltage.
I had the 6800 speedster. Heavy coil whine and started artificing. Send for RMA, after a week, I'm being told that there's nothing wrong with the card. Game again and starts doing the same thing. Heavy sweat around the pads too, everything covered in oil. Send it back with video evidence and literally 4 days later same story. Ended up selling the damn thing. Also crashed a lot but was really awesome fast. Upgrading from a GTX1650 super, that card was honestly great in performance -when it worked, but I loved the 2080Ti I got after this raw deal. Sold it at the beginning of last year and around June I got a 3080Ti. Honestly, aint that great of a difference between the 3080 12Gb I was borrowing from my friend and the 3080Ti I got now....
Just unhappy with this card cause it never exceeds 310w. My 2080Ti sometimes exceeded 400w, both on water. Feel like I can get more out of it but not worth the risk of losing it
Dude, I've watching your videos and kudos to you sir. You work with GPUs the same way I look into whole systems. Leaving no stone unturned. Having said that, I hope you can give me some idea as to what couldve happen to a GPU I have. The GPU is an evga rtx 2060, 2 years old. Purchased NEW. Never used heavily (only "old games") and in pretty good condition overall. There was a power outage and the PC was turned off (obviously) no UPS but a basic stabilizer only. When the power came back the PC turned on, nothing out of the ordinary. Went into a game and as soon as the map loaded the whole PC went kaput. Dead. After that, the PC wont post (1 long beep 2 short ones. VGA error) EXCEPT it does SOMETIMES. And when it does. i can run heaven benchmark, play games, stress test the GPU. No artifacts, no black screen no nothing..... But then again, SOMETIMES it goes to a black screen and the card seems to "dissapear" (I can hear the sound windows plays when a peripheral is disconnected, and if I switch to the iGPU I can see theres no GPU "connected") I know its nigh IMPOSSIBLE to say what could be wrong with it over a comment. But do you have ANY IDEA as to what could've caused that behavior? I mean a dead GPU is dead. But what about one that SOMETIMES works and when it does it SEEMS to be perfectly fine. Yet sometimes doesnt even get recognized by the system... I would really appreciate any guidance on where to start (electrical, memory, core) cause I have literally no idea. Sorry for the long post. TL;DR: Card boots SOMETIMES and works FINE. Sometimes it doesnt boot at all. And sometimes gets "disconnected" randomly and dissapears from device manager. No artifacts or nothing. Thanks!
Guess I win silicon lottery with my 6800XT. 3.5 years and no major issues, just one time this year where the pre-applied thermal paste dried up and the GPU hotspot jumped to 100+ °C and games running Unreal Engine would crash. Easily fixed with new thermal paste.
Truth hurts but it is what it is. I've seen the exact scenarios you mentioned Tony. 2 6800s dying seemingly randomly on 2 rigs. One gigabyte (though tbh, it could've been just because it was a gigabyte) and one XFX. My own 6900, an XFX 619 ultra, went for RMA twice because of a dead board.
Have you seen issues with the 7900xt/xtx? i am looking at buying one soon. Wanting something newer than my 3080, but not liking the 4000 series and 5000 likely will be worse. edit: oh boy heard what you said in the video, do you have any board brands you have seen that are LESS crap lol
Damn, throwing shade at my beloved Maxwell, my 980ti got me from 2015 up until...actually it only lasted till 2018 before one of the HDMI ports started having issues and i had to RMA it to EVGA...got a recertified 1070 instead...but anyway, 980ti, loved it.
A lot of these have terrible hotspots that amd says fine but its really not if you want it to last past warranty. Some of these have like 30-40 c gap between average and hotspot, if your gpu is like that its cooked, quite literally in this case.
Please let us know what you think of 5090 power requirements and of the card when you get one. I have not upgraded for 5 years now but fairly afraid to consider it due to the said power consumption
Not related to the problem faced in this video, but I bought the Reference RX 6900 XT on release, one of the worst experiences only second to an XFX RX 5700 XT that I resold in days after buying. Kept the RX 6900 XT for years, even though the Core reached 90°C+ and the hotspot reached 113°C+ since early on. Wasn't until the HDMI and one DisplayPort died that I finally RMA'd it. They sent back a refurbished ASUS TUF model and the experience was night and day, temps were amazing (barely reached 70°C and hotspot 90°C). In my case, guess the Reference cooler was just dogshit and caused crashes and the ports to die overtime, otherwise I don't know what happened since the cables and monitor still work.
"The worst GPU AMD ever made." Me with my Gigabyte 6800XT OC and all the issues I been having with it.👀 About to send it back to gigabyte while I still have warranty on it.
It would be interesting to get information from the repair customers about their type of usage. Is there something unusual and extreme about the conditions in which most failed cards have worked, or will these inherent problems arise in any conditions?
"900 series was a steamer" No kidding with all the reference 980 Ti's that have bad VRMs or memory chips and whatever the hell not that causes the card to power trip the whole system under load.
@@DJDocsVideos We were ahead but Canada and NA are making strides to catch up, although if it weren't for the EU regulations I highly doubt the latest iPhone would have achieved a 7/10 reparability score from iFixit.