HI, I'm the author of that post on Reddit. I was suffering from 110c hotspot, literally 1-2 minutes after starting a game, switched temporarily to a different monitor (HDMI 2.1) and noticed I no longer had the hotspot issue, was high 80s Low 90s. Strange, so I switched back to my display port monitor, 110c hotspot again after like 2 minutes. So I changed the DP cable for another I had purchased from amazon and temps were normal, just like when using the HDMI 2.1 monitor. So I cut open the problematic cable to check if it had 20 pins, it did but only 19 were physically connected. That was a few days ago, but now I do see temps creeping up to 110c again (new cable and HDMI 2.1) after around 10-15 minutes, so I do think I have a faulty heat sink. Now the golden question, why did I have almost instant 110c temps with that first cable, well after chatting with an AMD employee on reddit, he thinks the original cable could have had a defect "Most likely, bad cable = higher link training settings = more power. If I had to guess. That file can confirm it if you had both cables plugged in at some point recently", I sent him a log file but didn't get analysis back from him yet.
Hi, I have only one question to this Cable-solution: Why do AIB-Partner models of those graphics cards do not have the same temp problems? I do not understand how this is only an issue for MBA-Cards (AMD Reference Cards) if its a cable related issue...
@@NatureBoy2315 The same way most reference card do not have those issues as well. There's a reason why it is called a defect, and I made this video just to state that the cables MAY be doing something, but doesn't mean other issues aren't there in some cards
I am waiting for more test and infos about the display cable used. Maybe theses people switched to 4k/60Hz cables thus caping the fps generated by the card, I don't know
Der8auer made a video proving that the vapor chamber of the gpus is the problem: th-cam.com/video/26Lxydc-3K8/w-d-xo.html Edit: "prove" is a little exaggerated but it's very likely, that that's the problem.
Find a bad cable with pin 20 connected and see if you can reproduce the temperature problem. Chances are you have one if you've kept included cables from a lot of monitors over the years.
Not sure about DP but with HDMI the different bandwith and refresh options would be the best available between the monitor, cable and graphics card. There are relatively obscure parameters such as blanking time that have a big impact on the 7900XTX idle power consumption and VRAM frequency (as seen on the video "RX 7900 XTX High Idle Power / Stuck VRAM Clock Fix (CRU)" from Jo3yization. So basically changing to a better (or just different) cable also changes dozens of parameters.
I can see signal reflection or impedance issues being caused by bad cables. However how that would reflect into the actual graphics core and influence junction temperatures I don't immediately know...
When it comes to computer engineering, there's probably like a crap ton of things that consumers overlook, and the bad or negative aspect we focus on too much.
@AncientGameplays yes lol. After a yr and changing paste over the yr it made the really soft pads crappy. So I replaced those with the gray puddy like pads. That's when I noticed instantly the 110c. Today I'll get the artic pads. The proper soft ones. Like blue/green ones.
But how is der8auer have better temps depending on the alignment of the card? (Vertically / Horizontally) He did not swapped the cable in the tests. Then at least there are two different Problems with the Hotspot temps or am i wrong with this?
If you ask me all I can imagine is a cheap cable overheating, changing voltage values as resistance increase and forcing the power controller on the GPU to swap into PCI-E back and forward and increasing power spikes and so on...
This is interesting. I saw a few videos about it and it seems like this is only reference card problem, but there isn't stated what percentage of those cards are faulty. This is strange problem and I think this is probably due to some combination of small issues. Maybe some batches of cards have less cooling liquid in vapor chamber, maybe bad cables are causing card to try to pump too much power, which converts into too much heat to dissipate? This is such an controversial generation of cards with all kind of issues. We have the biggest price increase in years, problems with heat (overheating AMD, melting NVidia), too many promises and not many deliveries, beta drivers (AMD and Intel), low number of supplies (I think they wanted to sell out previous generation instead of new one). I hope this year will be better and we will see improvement in price and performance of the cards, especially in mid to mid-high range models.
Feels like unterminated end on a cable leads a leak in gpu's output(port) section and OG version doesn't have a vapor chamber to support such a leak while others can. This means amd didn't test it with such cables and put a proper solution on pcb or find a better way to implement output section in gpu.
I had a 5700 XT running at 110°C junction temp immediately after starting a game, but I solved changing the thermal pads. Could the causes have been multiple?
Something I notice with Vapor Chamber designs in general, AMD's Vapor Chambers are designed to be too big and wide over the entire PCB, perhaps multiple heat sources messed with how the gases flowed when the fluid evaporates, so if the entire board is quite hot or warm, the gas flows and is trapped in a area that isn't efficient? The faster tech VRAM we get, the hotter they get, and at low pressure, fluid vaporizes more often, and from multiple sources. Sapphire Nitro+ 7900 XTX uses Vapor-X, which is like 40-60% smaller in size than AMD's design, but makes use of heatpipe to heatsink fin stacks to spread and help dissipate heat. Vapor Chamber alone isn't sufficient, and you also should know that most mid to high end Nvidia GPUs do use Vapor Chamber, as this component offers up to a 30% efficiency over regular heatsinks. Only AMD Reference cards in the past uses Vapor Chambers, and it's often in a blower style hence very loud. Most AIB cards in the past other than the highest or the most expensive types, uses traditional heatsinks for AMD GPUs. So AMD gained a reputation for 'hot' temps because they never attempted to copy Nvidia's use of the more expensive, and custom vapor chambers at their mid range prices (I'd assume the extra cost is hefty for vapor chambers).
Maybe it has something to do with that Compression people spoke about with NVidia and DP1.4(or whatever it is they use on the 4000Series). If the GPU has to compress the Picture to fit through the Cable, maybe a very specific part of the GPU gets used for that all the time=110°C. Just speculating!
Display cables matter a lot. Years ago when I built a PC, the PC would not boot (froze in bios boot screen) Pretty sure it was the 20th pin causing some problems. I bought the cable from Canada Computers and all their display cables which they brand as iCAN caused this issue ( the employees tried in the store to boot my PC and they had the same issue I had ). Later on I bought a 3DClub displayport 1.4 cable and the issue went away.
Question: My GPU (sapphire 7900xtx) survives relatively well a time spy stress test (94ºc junction / 70 edge) with default settings. But it finds the 110ºc junction temp when doing an auto OC at ~380 W, with a slight downthrottle to ~365W during the 20 loops, enough to fail the stress test. Is this a RMA situation or just a sour bad luck on silicon lottery? I am almost wishing for this to be a RMA case as to get a new XTX, but I feel it is different from the whole hotspot situation.
I can confirm this with a 7900xtx hellhound powercolors, changed the cable with the original of the monitor and the temperature went down 10 degree celsius at extreme undervolt performance, and even more in iddle.
Would you know how to fix the "pc freeze" issue we get on some games like lost ark for rx6000 series card users? It does not happens on all games but several games I play is affected. It's like a random screen freeze after playing a bit that locks up the pc and force you to hard reboot. (I read that there's a high chance of driver issue.) I have a rx6800xt
@@AncientGameplays perhaps it's that? Tho i have run benchmarks for hours without issue so dumno... The more I look for solution online the more ppl I find with that kind of issue with random games, all using AMD cards... If you have time, can you try lost ark with your Rx 6800xt? (If you don't have any issue and using an old driver, please tell me which version you are using^^)
The vapor chamber issue has been proven for some users since the problems gets resolved when changing from horizontal to vertical and seems to be a common issue. The cable-fix I think is the issue for a very small minority of users, just like the bad mounting og cable og 4090 was.
I agree with the mounting. I’m running the Rx 7900 xt (not xtx) and noticed it was very hot while mounted horizontal. The heat, which rises, had almost no where to go. Got a vertical mount that also kept it away the glass on the case, and the entire case seems to be cooler. I didn’t check the actual temps to compare but with the hot air able to get out of the heatsink at the top and the fans getting better airflow everything seems cooler.
Perhaps the GPU works too hard with certain cables, because it's trying to funnel extra 'data' to the monitor for whatever reason, to no change in visual fidelity than frame rates? Also consider the monitor itself, does it offer high refresh rate? If it's a 60 hz monitor it shouldn't be recieving 90~120 fps signals... You won't see anything.
I just ordered a new monitor, and ordered a 7900 XTX weeks ago, so now I'm paranoid and looking for a good cable but couldn't find any brand I recognize selling a DP 2.0 cable on Amazon but Monoprice, so I ordered that for now. Any suggestions?
My 7900XT runs fine as I've been testing it daily since I got it 5 days ago. The hotspot or junction hits a max 82c no matter what I do. I have the reference PowerColor
My xtx card(reference) when default dont go above 84c...but if i increase the power limit or apply overclock that is when the card goes to 100c and so on, in junction...
I just had a problem with a monoprice dp cable and the #20 pin being connected. All my pc would do was stay in a boot loop but would be fine with hdmi. After I shut my pc off and unplug it the cpu light would stay on as long as the monitor was plugged in and connected to the gpu
Another thought on the hotspot temp versus GPU temp was the differential increasing. You would think if the vapour chamber failed then the temps would increase temp at a similar dT. With the cable it's surprising that there isn't some current limiting because a bad cable would overheat if it's impedance is incorrect. To much speculation...
I bought a XFX 7900XTX reference card at launch, I never checked the junction temp, but it has really noticeable coil whine. Is that a side effect from high temps? I might have to test my temps when I get back from vacation. If my card is affected, can I hit up XFX support and ask for a replacement? They probably don’t have anymore reference cards in stock but I wouldn’t mind getting a MERC310 7900XTX.
I've got a 7900 xtx red devil. I have about 65-75c on the GPU itself, but 98-105 junction sometimes 110. However it does not throttle, and the fans stay around 1300RPM. Should I worry?
@@AncientGameplays I've got a Corsair 4000X case with 3 intake 2 top and 1 exhaust fan with a 360mm AIO in front of the 3 intake fans. If I run my case fans around 1700-1800 rpm I get 99-105 consistently while playing rdr2. Should I contact powercolor about it or that's fine? Truth be said the GPU is really crammed in there just barely fit in my case, I had a 3080 prior to this GPU and I never checked junction temps but GPU core temps are similiar to my old GPU.
you should at least increase the fan speed in adrenalin software.. max 60% at least .. will improve a bit without too much noise. Otherwise, best thing is to repaste the card which is described in other videos.. I have the same red devil issues. i'm enjoying it with faster spinning fans. havent repasted yet. still works great but junction often 95°
Hoverheating oooooooooo i hate this problem i have gygabyte gaming oc Rx 6750xt and is very hot so i put new settings and i change the thermal past (thermal grizzly Kryaunot) I also made a ventilation curve and with all this I stay in the 70 to 80 degrees or 85 maxxxxx but what I don't understand is that it heats up more in the cinematics than in the gameplay would you have an idea why?
I recently bought a Ryzen 5 7600x with a 240 mm Lian Li Aio, but the temperature of the processor is terribly high, both during the game and in the Windows environment. In the game, the processor temperature reaches up to 95 degrees Celsius and in the Windows environment it is around 45-55. I mean, why is the temperature so high even with this Aio ? Is there a solution?
It is not high at all my friend. Those CPUs are made to boost up to 95C with absolutely no throttling, it's how they work. Still, you have something wrong there as even my 7700X barely reaches 75C on heavy CPU gaming with a noctua DH15-s. Check you case airflow
Side note/ related question? I've had an RTX 2060 for some time now. But, can't seem to make the correct power draw numbers appear. I've tried driver updates and NVCleanInstalls. I've re-installed MSI Afterburner twice now. Checked X,Y, Z setting and tick box. Re-installed HWiNFO64, checked for it's readouts... Nothing there either. Could my card be faulty, or is there some stupid trick I just haven't tried yet?
What I don't understand about identifying vapour chamber as the problem is that no one tried swapping in a waterblock to verify that this was the issue. An easy enough verification. Also, someone could order a dozen different DP 2.1 cables and see if there is a correlation. The problem with some tech tuber's (not you) is that it's about the conspiracy, and chasing the YT algorithm, and not about accuracy.
der8auer has done some decent legwork on this, but has definitely not taken it to completion. The youtube tech channels and the youtube linus's will be all over this, give the fanboys what they want, make the $$$
@@TorqueKMA No i ordered an EK water block on the third day of the 7900XTX release, I will be getting the block around mid to late February that's what my pre-order says.
Fabio tell me how can you see all the detailed temps, do you have an iCUE Corsair case? Along with this software or MSI Afterburner? And MSI Afterburner Cannot run by itself without the hardware help of something like Corsair iCUE Case right? I apologize for being noob master
@@AncientGameplays And if I use the MSI Afterburner while doing some Furmark test and some gaming and see these high temps over 80-90 degrees, should I return the card ? It's something wrong with the pcb or heatsink? I m talking about 7900 cards. 7900xtx is between 100 and 200 dollars more expensive, do you think it's worth it for what it gives along a Ryzen 7600? Also, is 7600 Really that much better than the freaking 5800X3D in gaming like benchmarks show on youtube and internet? But why? even the Non X version with much lower frequency and smaller wattage? Is 7900xt better than the XTX variant having better temps and lower noise levels? Because its kinda weird to spend 100-200 euro or dollars on a card that gives you worse temps, a lot of noise, bigger power consumption, and the sake of some drunken 10 filthy fps... also, is 6800 (not XT) < which is like 40 dollars more expensive than 6750xt> better as a price/performance ratio? I know I m complicated and I deeply apologize my friend, I don't want to create problems I swear, just wanna clear dilemmas out of my stormy head and have a nice gaming rig.
When I got my reference 7900 XT on launch day I also saw 110c junction temp after just a few minutes. But just today I ran speed way benchmark for 30 minutes and hot spot didn't go above 82c with gpu temp at 65c, strange. Perhaps the latest driver I downloaded fixed the issue, I advise everyone to update your drivers asap.
after watching your last video it got me thinking and i had the idea that if you are spending that kind of money on a GPU why wouldnt you just put a water block on it? or is a vapor chamber better for cooling?
I think cables have been an issue for a while now and its really easy to get a bad cable even if you would spend a little more on a way better one. Would be surprised if this is the issue but i dont see it helping i having a bad cable
Hi, do you have problems as well with the hardware acceleration in browser with AMD gpus? I have RX 6800 and while watching YT on Mozilla/Chrome/Brave videos stutters/glitches etc. Do you have some solution for it? Thank you!
My reference XTX runs closer to 73-75c horizontal in mid tower ATX vs 70-72c vertical in a mini ITX. should I be concerned about the difference in temps? GPU is stock settings default mode.
Very interesting topic. What about changing the PCIE cables? i would like to test that (probably a waste of time) but I will definitely try to replace my display port cable to see any temp differences
Hi friend! sorry I have a problem with my new Rx 6650 xt I can no longer play 4k videos on TH-cam, before I had a gtx 1050 and if it allowed me! my monitor even being 1440p gave the option to watch videos at 4k! What could be limiting it?
How can anyone report responsibly that there wasn't enough fluid, when Debaurer said himself 'I didn't measure the fluid when I cut it open'. You CANNOT say 'a few mm's too little' from a few pictures of open mesh.
Vapor Chamber requires the fluid to turn into gas, meaning it can't have too much water/fluid like a liquid cooler, it's also a low pressure container meaning the fluid vaporizes at lower temperatures, so water wouldn't need 100 Celsius to boil over, but say 80 instead? As well, roughly 0.1% of all MBA reference cards might have this defect, because say whomever assembled it (usually a robot) ended up making a mistake.
I just done a re paste on a reference 7900xtx with the thermal right TFX and TF7 mixture, (very thick) with minimal pump out, reduced hotspot temps from 50c delta to 14C delta with a 10 hour stress test, also with thermal 4mm thick pads behind the core, 74c temp with 98C hotspot 420 watts 3018mhz core, memory is at 91C doesnt go higher than that, at UV its currently sitting at 67C 83C hotspot, 81C memory temps, success i guess (reference cooler btw)
these were some defective batches, you can't fix defects, the cards are send back and then the consumer gets a new one. The "Normal" cards never had such issue
@@AncientGameplays so if i buy a card today in this very moment would be a "normal" gpu? that's all i want to know.... the new batches coming to restore de stock are the fixed ones(normal)?
OP of the reddit thread just said the issue has returned after moving his PC case. I think Der Bauer has provided the reason for the hotspots along with Igor being that there is a lack of coolant.
Hi again Fábio. Looks like my reference XTX reaches the 110C with the power limit +15%, while stock it stays at about 90C. Do you consider this one as a faulty card?
thanks for the tips, i m actually in this situation (7900xtx) with some hotspot at 110° and higher, i will try to switch my HDMI cable to a DP port i guess and see if it resolve the issue
What worked best for me Fabio was cleaning out the thermal paste and then using a Thermal Grizzly Kryosheet, this stuff is magic! I saw many people referring to the PTM7950 however I could not access this in my country, I managed to get the Kryosheet and my Hotspot and GPU delta dropped from around 50 degrees celsius to generally under 20 degrees celsius. Only down side in my opinion is that it is electrically conductive so you have to be careful when putting it on the die. I figured my problem on the 7900XT was due to pump out as thermal paste only improved temps for about a week before they got worse again. If PTM7950 is availabe it might be a better option as it does not conduct electricity however both are great products!
I was getting bad temps, crashing with 3 cables rated for 8K/hdmi 2.1 and wasn’t gaming in but 4k or 1440p. I also had a very thin, but from top to bottom pixelated verticals line. I made sure bios Was up to date. I removed Nvidia drivers. Factory reset the 7900 XTX MERC10 4x and my fix came with a Nintendo Switch HDMI. Which pulls 120hz. Gpu usage and temps came down dramatically. I can even play with overclock settings and got 17599 on Port royal. Factory cooling. To tell ya how bad things got, it was like I had to treat this like a 1080P card coming from an Rtx 3080. I still have a 5900x. I also have a gaming x trio 7900 XTX. Which has had zero issues. Merc10 hasn’t had issues for awhile since it went away. Another possibility is that the games in combination have been messing with the GPU. Like I was playing Fortnite and was messing around in settings on 1440p, getting a feel for the ray tracing in addition to handling the most basics on epic with tsr epic. I still had around 58-62% gpu utilization. I believe my run through the situation, includes playing Fortnite with ray tracing on while having the Radeon overlays on to view temps. Nothing seemed out of whack until it was. Zero over clocks performed, or memory, undergoing or any performance tuning for that matter existed before the issue. Since we’ve had a Fortnite update. I still play with lumens and hardware ray tracing off. I’m not surprised that Fortnite can be demanding. I remember when Rtx 2070 super handled 1440P, then Rtx 3080 handled 1440P dx11 Fortnite. Now even with 4080 or 7900 XTX, the quality of the game can be so high that 4K is not well at native unless the rest of the settings are low.
It looks to me like one or more batches have no (t enough) fluid in the vapour chamber. This explains why it works vertically because the little amount of fluid there is has a chance to drop back down. On yours it doesn't matter because it has enough fluid. That's what I think anyway. When your sweet cooling solution fails because you forgot a drop of water. 😂
Wow, this is really surprising. I'm working on SPS power chips, not on GPUs themselves. And in my limited knowledge there isn't a way that the IO part of the GPU should influence the core temperatures, but seeing as it's not an isolated case. I'll keep my eyes open to further developments
So the thing is, it's not the core temp that's high, it's junction temp. Junction sensors are a network of sensors embedded in the die, the AMD software is either displaying one of many or one particular one, so it's actually possible that the one being displayed, or the hottest one is localized in one part of the chip that's doing one specific thing.
@@PineyJustice In my head I imagined IO to not be part of the die and was a separated chip, but thinking about it and considering the huge bandwidths we're having these days I guess it'd make sense to actually be on the die itself. Still quite an odd behavior. It'd be interesting to see if connecting that Dell cable (or any with pin 20 wired...) to an Nvidia card or a previous Gen AMD would result in any change in temperature.
Wow. I was thinking i was gonna have to repaste my gpu because my hotspots almost stayed at 110º. Swapped my dell alienware 4k monitor from a dp to a hdmi and my hotspot went from pegged at 110 down to 80-90 during gaming.
I hope der8auer transplants a known-good vapor chamber (like yours) to one of his cards that is overheating to see if that fixes the issue. If the issue is resolved after swapping the vapor chamber, then we know it is indeed the chamber, but if the issue persists then it may be something else.
You would not be able to tell by doing that though. As one person sent him a card that was thermal throttling. A change of thermal paste fixed the card. So when you swap the coolers. Thermal paste would be changed. Would it be the cooler or the thermal paste ???
@@suzie9874 He already excluded the thermal paste, though with a very small sample size of course. So I think it is less likely to be thermal paste and more likely the cooler or some other weird issue (as discussed in this video), or more likely a combination of both as the temperature changes with different orientation and different cables.
@@myname7021 I just seen that they also have an issue with higher heat with 20 pin monitor cables. When using 19 pin cables the temp would lower. Odd i know but i did see this somewhere.
I finally got some advice about my new 6700xt (used from a crypto miner). The person apparently had had the same exact problems. (bios constantly resetting and not restarting from bios without hitting the reset button). He seems to have thought I got a bad card so I had 20 days left of warranty and sent it back. They're going to send me a new card free of charge so I'm very happy with the outcome.
De momento ,los que intentan tramitar su grafica con el problema de sobrecalentamiento por RMA , no son aceptados ya que dicen que esas temperaturas son normales y no les tramitan la devolucion ni les dan una solucion .
Atomized Android and I told you, on your Discord, what the problem was and how to fix it. So let me repeat it here again. The working hypothesis is that it has to do with the very narrow fin pitch of the GPU cooler (the distance between the heat exchanging fins). This has been tested by Atomized Android a number of times on his 7900XTX and is repeatable, and during testing his GPU reaches 400 Watts TBP. When running the card horizontally, the narrowness of the fin pitch does not allow for the air to be ejected out of the with enough force (fans don't have enough Static Pressure?) and so the warm air is being sucked into the fans again and recycled. Atomized Android with his 7900XTX reference design card came across this by accident. He was using is previous 6900XT with a program called "Fan Control" which would ramp up the front intake fans of his case when both the CPU or the GPU came under load (unlike normally where you have to choose if either the CPU or the GPU controls the fan speed of the front intake case fans, but not both). When he put in his new 7900XTX the profile was no longer applied, because it didn't recognize the GPU (it was configured for a 6900XT). When he was running his tests, the GPU went up to about 108 degrees Celsius and he was a bit annoyed. He noticed however that the front fans in his case weren't ramping up the way they had before, although the fans on his GPU were going like the clappers. So he looked at his Fan Control profile, noticed the problem, configured it the way it was before for his 6900XT, ran the test again (this time with the front fans ramping up to blow air over the card when the GPU came under load) and his max temp at a 400 Watt TBP was at around 84 degrees and the delta between the "Current Temperature" (edge temperature) and the Junction Temperature was around 16 degrees. You showed Der8auer's efforts at getting to the cause of the problem and the only bloody thing he didn't do was push some airflow over the card with a fan when it was horizontal. This would explain why the problem is so random. It would be bloody easy for you to test this with your system. All you would have to do is have the intake fans of your case set to various speeds and see the difference it makes to the max temp your card reaches. Start with full speed and then ramp down until you find the point where your temp is optimal. Remember if you have your front case fans set to ramp up dependent on CPU load that games don't really stress the CPU and the better the cooling you have on your CPU the less likely your front case fans are to spinning up. So there you go, job done.
I brought a new cable i feel like its dp cable 1.4 from cable creation now my hotspot temps reach upto 100 degrees in an instant but it happens mostly in rdr2 for some reason
The Vapor Chamber is the problem not the cable, The issue is the 20 Pin cables are causes the port and controller to heat up, which interferes with the Vapor Chamber's wicking and return paths for the condensation. Putting the card horizontal causes heat to pool on the controller.
@@AncientGameplays Likely cause the cheap cable doesnt enables Sync Options, allows the GPU to run full tilt in gaming, where as the better cable syncs the GPU, and limits the rendering. Either way, There's cases and results that prove an Display Port cable isnt the problem.
@@AncientGameplays Power Draw fluctuates during rendering, these could be cherry picked screen grabs, even the order of the layout changes, last time I checked, changing a cable does not change my OSD Order and Layout. Pin 20 Problem was fixed 5 yrs ago by both AMD and nVidia Thru Driver updates. TO Prove a Cable would do this, there needs to be a Video, of the screen and card, showing when the cable is switched while running the bench mark, the temps drop. Shutting off the benchmark to switch cables allows the VC to cool and restore functionality. AMD Already announced they discovered Vapor Chamber defects where the VC would "Vapor Lock" and the condensed fluid would not return to the GPU plate to continue cooling. So the point is moot.
The display port on this generation is not the same, that alone should demand a next gen, certified cable. Also, the vapor chamber is something prone to all sorts of problems, why bother at all??
Thanks I decided to try and buy a 7900 xt merc (the highest boost model) for 900$ on newegg. I bought it but then later it said i needed to have an adult signature when it comes. Why would newegg take my money before they tell me that? Now i haft to wait 3-5 days to get my money back and also 4 months so i dont haft to go through that bullshit again.
tell a lie, with your max overclock settings junction temp hit 109c on furmark in about 2 mins, so turned off the benchmark, but guessing this is extreme and under normal settings ran it and it seemed fine
Cool man, yeah so all seems good, I did get a little panic on when I used your max oc settings, and then ran it again, shot up to the 110c before I switched off the benchmark, the temps really shot down quickly to normal which I was happy with. I ramped up the fans to 100% and with oc temps still shot up, took about a minute and a half to two mins to reach 110c but literally second to drop back down when turning off the benchmark. I since read in the furmark notes say not too stress test with it on oc settings, whoops! :-) but stock worked ok 😅 and all looks good for me, hoping not to many people suffer with the problem. Loving the card so far 😁
Der8auer really needs to open up a vapor chamber on a known working card to know for sure what is going on. But people aren't in the habit of sending away their working products to be destructively dissected....
In his video he talks about how he sent some 4080s & money if they did not want that as well if not one also said if you can send me a AIB model some would take that as well. So any card that was sent he paid for or replaced it for what they asked for. I think He got 3 shipped in if I remember right. I get some might be afraid but Der8auer never risk his name & not give the card back or something in return.
Nice video like always - for cable try whit your 7900xtx The PremiumTech Europe DisplayPort 2.0 Cable and whit last video of yours overcloking and undervolting 7000 cards temps shall be ok.
I got my 7900 xt merc black edition :), lowered voltage to 975mv instead of 1100mv i think im around my limit, i wish i could edit power play tables with mpt I would get way way higher clocks.
Not the case, I've known for YEARS that a good cable should always be a primary thing because it could cause issues like flickering, blackscreens and even booting problem. Its a known thing. like this though, I never saw, but since the new AMD cards use chiplet design and a new Display Engine, it may be possible
WTF this absolutely worked for me. I had a dell displayport and swapped it to a different one and went from 67/108 hot spot to 50s/low 90s. Im really curious what the actual issue here with this. edit - My card is an Asrock 7900xtx Taichi white. This fixed my heat in game but my card still sucks. It goes unstable the moment I undervolt even a little in games but I can undervolt it hard for benchmarks..rip
That's due to power scales and the games you play. Because benchmarks are usually heavier, meaning that the "base" voltage is higher since it is using a higher power state. In games, some can be use a higher base voltage and others a lower one, meaning that with the same slider option the voltages will be different. I myself only use down to 1130mv. Want to really go power saving? Decrease the frequency
@@AncientGameplays i have to sit at 1150 but interesting. I wish I knew this before spending a night running 3Dmark getting higher and higher scores while overclocking to have my night shattered by the first game lol.
HI, I'm the author of that post on Reddit. I was suffering from 110c hotspot, literally 1-2 minutes after starting a game, switched temporarily to a different monitor (HDMI 2.1) and noticed I no longer had the hotspot issue, was high 80s Low 90s. Strange, so I switched back to my display port monitor, 110c hotspot again after like 2 minutes. So I changed the DP cable for another I had purchased from amazon and temps were normal, just like when using the HDMI 2.1 monitor. So I cut open the problematic cable to check if it had 20 pins, it did but only 19 were physically connected. That was a few days ago, but now I do see temps creeping up to 110c again (new cable and HDMI 2.1) after around 10-15 minutes, so I do think I have a faulty heat sink. Now the golden question, why did I have almost instant 110c temps with that first cable, well after chatting with an AMD employee on reddit, he thinks the original cable could have had a defect "Most likely, bad cable = higher link training settings = more power. If I had to guess. That file can confirm it if you had both cables plugged in at some point recently", I sent him a log file but didn't get analysis back from him yet.
Maybe that first cable (being a dell cable and has no click when inserted) , maybe I had not connected it fully.
@@FTLN Still does not explain the temps, its literally just an output if there's not 20 pins.
Hi,
I have only one question to this Cable-solution: Why do AIB-Partner models of those graphics cards do not have the same temp problems? I do not understand how this is only an issue for MBA-Cards (AMD Reference Cards) if its a cable related issue...
@@NatureBoy2315 The same way most reference card do not have those issues as well. There's a reason why it is called a defect, and I made this video just to state that the cables MAY be doing something, but doesn't mean other issues aren't there in some cards
@@NatureBoy2315 the board partners may not be using a vapor chamber to cool the cards. Those AIBs don't have the issue because of that
As always hope you enjoy this info! What do you think about this?
I am waiting for more test and infos about the display cable used. Maybe theses people switched to 4k/60Hz cables thus caping the fps generated by the card, I don't know
Der8auer made a video proving that the vapor chamber of the gpus is the problem: th-cam.com/video/26Lxydc-3K8/w-d-xo.html
Edit: "prove" is a little exaggerated but it's very likely, that that's the problem.
Hahahahaha! Home boy cut his cable
I think it's good you don't get on the bandwagon
@@timo4257 If he 'proved' it was the problem, how come his latest video today says that vapor chamber wasn't?
Find a bad cable with pin 20 connected and see if you can reproduce the temperature problem. Chances are you have one if you've kept included cables from a lot of monitors over the years.
Not sure about DP but with HDMI the different bandwith and refresh options would be the best available between the monitor, cable and graphics card. There are relatively obscure parameters such as blanking time that have a big impact on the 7900XTX idle power consumption and VRAM frequency (as seen on the video "RX 7900 XTX High Idle Power / Stuck VRAM Clock Fix (CRU)" from Jo3yization. So basically changing to a better (or just different) cable also changes dozens of parameters.
I can see signal reflection or impedance issues being caused by bad cables. However how that would reflect into the actual graphics core and influence junction temperatures I don't immediately know...
Same here
I love how quickly you get on this stuff. Very interesting that those cables are making such a difference.
When it comes to computer engineering, there's probably like a crap ton of things that consumers overlook, and the bad or negative aspect we focus on too much.
I just bought the club3d cable. Didnt fix it still 110c
You have a defective card maybe friend...
@AncientGameplays nah its just the pads. Today new proper pads arrive. I'll update later after work.
@@rd2dab obviously defective though. You had to change pads...
@AncientGameplays yes lol. After a yr and changing paste over the yr it made the really soft pads crappy. So I replaced those with the gray puddy like pads. That's when I noticed instantly the 110c. Today I'll get the artic pads. The proper soft ones. Like blue/green ones.
@@rd2dab and results ? Hotspot temperatures now ?
Crazy was hitting 112 junction now I’m at 84 🎉 haven’t tried my second screen yet gpu only has 1 hdmi port
Great!
But how is der8auer have better temps depending on the alignment of the card? (Vertically / Horizontally)
He did not swapped the cable in the tests.
Then at least there are two different Problems with the Hotspot temps or am i wrong with this?
He didn't
Because these are untelated issues
If you ask me all I can imagine is a cheap cable overheating, changing voltage values as resistance increase and forcing the power controller on the GPU to swap into PCI-E back and forward and increasing power spikes and so on...
This is interesting. I saw a few videos about it and it seems like this is only reference card problem, but there isn't stated what percentage of those cards are faulty. This is strange problem and I think this is probably due to some combination of small issues. Maybe some batches of cards have less cooling liquid in vapor chamber, maybe bad cables are causing card to try to pump too much power, which converts into too much heat to dissipate?
This is such an controversial generation of cards with all kind of issues. We have the biggest price increase in years, problems with heat (overheating AMD, melting NVidia), too many promises and not many deliveries, beta drivers (AMD and Intel), low number of supplies (I think they wanted to sell out previous generation instead of new one). I hope this year will be better and we will see improvement in price and performance of the cards, especially in mid to mid-high range models.
Feels like unterminated end on a cable leads a leak in gpu's output(port) section and OG version doesn't have a vapor chamber to support such a leak while others can. This means amd didn't test it with such cables and put a proper solution on pcb or find a better way to implement output section in gpu.
I'm glad you're doing all this research and testing so I won't have to once I get mine.
It just came to my attention
Gonna buy it just in time for the recall
I had a 5700 XT running at 110°C junction temp immediately after starting a game, but I solved changing the thermal pads. Could the causes have been multiple?
Yeah, for sure they were
Something I notice with Vapor Chamber designs in general, AMD's Vapor Chambers are designed to be too big and wide over the entire PCB, perhaps multiple heat sources messed with how the gases flowed when the fluid evaporates, so if the entire board is quite hot or warm, the gas flows and is trapped in a area that isn't efficient?
The faster tech VRAM we get, the hotter they get, and at low pressure, fluid vaporizes more often, and from multiple sources.
Sapphire Nitro+ 7900 XTX uses Vapor-X, which is like 40-60% smaller in size than AMD's design, but makes use of heatpipe to heatsink fin stacks to spread and help dissipate heat.
Vapor Chamber alone isn't sufficient, and you also should know that most mid to high end Nvidia GPUs do use Vapor Chamber, as this component offers up to a 30% efficiency over regular heatsinks.
Only AMD Reference cards in the past uses Vapor Chambers, and it's often in a blower style hence very loud. Most AIB cards in the past other than the highest or the most expensive types, uses traditional heatsinks for AMD GPUs.
So AMD gained a reputation for 'hot' temps because they never attempted to copy Nvidia's use of the more expensive, and custom vapor chambers at their mid range prices (I'd assume the extra cost is hefty for vapor chambers).
that may be an issue. Also, heard that the case airflow matters a lot in there
Maybe it has something to do with that Compression people spoke about with NVidia and DP1.4(or whatever it is they use on the 4000Series).
If the GPU has to compress the Picture to fit through the Cable, maybe a very specific part of the GPU gets used for that all the time=110°C.
Just speculating!
Display cables matter a lot. Years ago when I built a PC, the PC would not boot (froze in bios boot screen) Pretty sure it was the 20th pin causing some problems. I bought the cable from Canada Computers and all their display cables which they brand as iCAN caused this issue ( the employees tried in the store to boot my PC and they had the same issue I had ). Later on I bought a 3DClub displayport 1.4 cable and the issue went away.
Question: My GPU (sapphire 7900xtx) survives relatively well a time spy stress test (94ºc junction / 70 edge) with default settings.
But it finds the 110ºc junction temp when doing an auto OC at ~380 W, with a slight downthrottle to ~365W during the 20 loops, enough to fail the stress test.
Is this a RMA situation or just a sour bad luck on silicon lottery? I am almost wishing for this to be a RMA case as to get a new XTX, but I feel it is different from the whole hotspot situation.
Im your case I wouldn't bother. Just tweak the card and get better case airflow 💪
I can confirm this with a 7900xtx hellhound powercolors, changed the cable with the original of the monitor and the temperature went down 10 degree celsius at extreme undervolt performance, and even more in iddle.
Interesting
Would you know how to fix the "pc freeze" issue we get on some games like lost ark for rx6000 series card users?
It does not happens on all games but several games I play is affected. It's like a random screen freeze after playing a bit that locks up the pc and force you to hard reboot.
(I read that there's a high chance of driver issue.)
I have a rx6800xt
I sense a cpu/ram instability there
@@AncientGameplays perhaps it's that? Tho i have run benchmarks for hours without issue so dumno...
The more I look for solution online the more ppl I find with that kind of issue with random games, all using AMD cards...
If you have time, can you try lost ark with your Rx 6800xt? (If you don't have any issue and using an old driver, please tell me which version you are using^^)
The vapor chamber issue has been proven for some users since the problems gets resolved when changing from horizontal to vertical and seems to be a common issue. The cable-fix I think is the issue for a very small minority of users, just like the bad mounting og cable og 4090 was.
I agree with the mounting. I’m running the Rx 7900 xt (not xtx) and noticed it was very hot while mounted horizontal. The heat, which rises, had almost no where to go. Got a vertical mount that also kept it away the glass on the case, and the entire case seems to be cooler. I didn’t check the actual temps to compare but with the hot air able to get out of the heatsink at the top and the fans getting better airflow everything seems cooler.
@@bsweeney7086 if horizontal had it close to the glass that may also affect temps. Glad it help change mouting.
Perhaps the GPU works too hard with certain cables, because it's trying to funnel extra 'data' to the monitor for whatever reason, to no change in visual fidelity than frame rates?
Also consider the monitor itself, does it offer high refresh rate? If it's a 60 hz monitor it shouldn't be recieving 90~120 fps signals... You won't see anything.
I just ordered a new monitor, and ordered a 7900 XTX weeks ago, so now I'm paranoid and looking for a good cable but couldn't find any brand I recognize selling a DP 2.0 cable on Amazon but Monoprice, so I ordered that for now. Any suggestions?
You will most likely be fine don't worry
did they lean the machine forward or sideways while changing the cable? fix might only be temporary.
Nope, they didn't
Will new drivers be able fix the DP/Hdmi?
only if the issue is drivers to start with. This seems like a real issue with bad cables
My 7900XT runs fine as I've been testing it daily since I got it 5 days ago. The hotspot or junction hits a max 82c no matter what I do. I have the reference PowerColor
same here
Mine is at 68 72 degrees at the highest
I also have the thermaltake open case
I have the reference 7900 XT from AMD, junction max is 85c.
My card seems fine as well. GPU Hotspot temp has hit max 81. gpu temp max 67. But... the memory junction temp doesn't go below 50... maxed out at 88.
power color and I have the issue 96 & 110
Love your memes at the intros, love your charisma.
Your videos are great! 👍
Thanks :D
My xtx card(reference) when default dont go above 84c...but if i increase the power limit or apply overclock that is when the card goes to 100c and so on, in junction...
100C is "okay", 110C is not. With tweaking it goes below 100C
okay is it possible that a new dispaly cabel can fix5700xt also or does it only work on 7900xtx
A good cable is ALWAYS needed to avoid issues
I just had a problem with a monoprice dp cable and the #20 pin being connected. All my pc would do was stay in a boot loop but would be fine with hdmi. After I shut my pc off and unplug it the cpu light would stay on as long as the monitor was plugged in and connected to the gpu
Another thought on the hotspot temp versus GPU temp was the differential increasing.
You would think if the vapour chamber failed then the temps would increase temp at a similar dT. With the cable it's surprising that there isn't some current limiting because a bad cable would overheat if it's impedance is incorrect. To much speculation...
I bought a XFX 7900XTX reference card at launch, I never checked the junction temp, but it has really noticeable coil whine. Is that a side effect from high temps? I might have to test my temps when I get back from vacation. If my card is affected, can I hit up XFX support and ask for a replacement? They probably don’t have anymore reference cards in stock but I wouldn’t mind getting a MERC310 7900XTX.
Coil whine has nothing to do with temps, its just electric vibration. It goes away or gets better if you undervolt
I've got a 7900 xtx red devil. I have about 65-75c on the GPU itself, but 98-105 junction sometimes 110. However it does not throttle, and the fans stay around 1300RPM. Should I worry?
105C is okay, 110C not really... thats very odd even more on a red devil version
@@AncientGameplays I've got a Corsair 4000X case with 3 intake 2 top and 1 exhaust fan with a 360mm AIO in front of the 3 intake fans. If I run my case fans around 1700-1800 rpm I get 99-105 consistently while playing rdr2. Should I contact powercolor about it or that's fine? Truth be said the GPU is really crammed in there just barely fit in my case, I had a 3080 prior to this GPU and I never checked junction temps but GPU core temps are similiar to my old GPU.
you should at least increase the fan speed in adrenalin software.. max 60% at least .. will improve a bit without too much noise.
Otherwise, best thing is to repaste the card which is described in other videos..
I have the same red devil issues. i'm enjoying it with faster spinning fans. havent repasted yet. still works great but junction often 95°
@@incakola486 i fixed it by buying a 4080 lol
Hoverheating oooooooooo i hate this problem i have gygabyte gaming oc Rx 6750xt and is very hot so i put new settings and i change the thermal past (thermal grizzly Kryaunot) I also made a ventilation curve and with all this I stay in the 70 to 80 degrees or 85 maxxxxx but what I don't understand is that it heats up more in the cinematics than in the gameplay would you have an idea why?
The problem is not AMD there lol, its gigabyte. Apart from Aorus, they're cooling solutions aren't that great
😂thanks, Soon the 100k very beautiful course I hope you will go even further than 100k❤
I recently bought a Ryzen 5 7600x with a 240 mm Lian Li Aio, but the temperature of the processor is terribly high, both during the game and in the Windows environment. In the game, the processor temperature reaches up to 95 degrees Celsius and in the Windows environment it is around 45-55. I mean, why is the temperature so high even with this Aio ? Is there a solution?
It is not high at all my friend. Those CPUs are made to boost up to 95C with absolutely no throttling, it's how they work. Still, you have something wrong there as even my 7700X barely reaches 75C on heavy CPU gaming with a noctua DH15-s. Check you case airflow
Side note/ related question? I've had an RTX 2060 for some time now. But, can't seem to make the correct power draw numbers appear. I've tried driver updates and NVCleanInstalls. I've re-installed MSI Afterburner twice now. Checked X,Y, Z setting and tick box. Re-installed HWiNFO64, checked for it's readouts... Nothing there either. Could my card be faulty, or is there some stupid trick I just haven't tried yet?
What I don't understand about identifying vapour chamber as the problem is that no one tried swapping in a waterblock to verify that this was the issue. An easy enough verification. Also, someone could order a dozen different DP 2.1 cables and see if there is a correlation. The problem with some tech tuber's (not you) is that it's about the conspiracy, and chasing the YT algorithm, and not about accuracy.
Can you buy one for xtx yet?
der8auer has done some decent legwork on this, but has definitely not taken it to completion. The youtube tech channels and the youtube linus's will be all over this, give the fanboys what they want, make the $$$
@@-opus the 4090 boys are coming out of the woodwork to tell us how smart they were for their purchase.
@@TorqueKMA EK has one on their website. I'm sure that EK would be happy send derBauer one.
@@TorqueKMA No i ordered an EK water block on the third day of the 7900XTX release, I will be getting the block around mid to late February that's what my pre-order says.
Fabio tell me how can you see all the detailed temps, do you have an iCUE Corsair case? Along with this software or MSI Afterburner? And MSI Afterburner Cannot run by itself without the hardware help of something like Corsair iCUE Case right? I apologize for being noob master
Forget icue it sucks! Use msi afterburner or hwinfo64
@@AncientGameplays And if I use the MSI Afterburner while doing some Furmark test and some gaming and see these high temps over 80-90 degrees, should I return the card ? It's something wrong with the pcb or heatsink? I m talking about 7900 cards. 7900xtx is between 100 and 200 dollars more expensive, do you think it's worth it for what it gives along a Ryzen 7600? Also, is 7600 Really that much better than the freaking 5800X3D in gaming like benchmarks show on youtube and internet? But why? even the Non X version with much lower frequency and smaller wattage?
Is 7900xt better than the XTX variant having better temps and lower noise levels? Because its kinda weird to spend 100-200 euro or dollars on a card that gives you worse temps, a lot of noise, bigger power consumption, and the sake of some drunken 10 filthy fps... also, is 6800 (not XT) < which is like 40 dollars more expensive than 6750xt> better as a price/performance ratio? I know I m complicated and I deeply apologize my friend, I don't want to create problems I swear, just wanna clear dilemmas out of my stormy head and have a nice gaming rig.
When I got my reference 7900 XT on launch day I also saw 110c junction temp after just a few minutes. But just today I ran speed way benchmark for 30 minutes and hot spot didn't go above 82c with gpu temp at 65c, strange. Perhaps the latest driver I downloaded fixed the issue, I advise everyone to update your drivers asap.
Strange indeed
after watching your last video it got me thinking and i had the idea that if you are spending that kind of money on a GPU why wouldnt you just put a water block on it? or is a vapor chamber better for cooling?
that's for air cooling
Interesting, posted your video in the overclockers UK forum in the owners thread.
Thanks
Do these issues affect the 7900 XT model as well?
These are issues that can happen not for all cards though
I think cables have been an issue for a while now and its really easy to get a bad cable even if you would spend a little more on a way better one. Would be surprised if this is the issue but i dont see it helping i having a bad cable
Hi, do you have problems as well with the hardware acceleration in browser with AMD gpus? I have RX 6800 and while watching YT on Mozilla/Chrome/Brave videos stutters/glitches etc. Do you have some solution for it? Thank you!
Nope, but I use mozila Firefox
Great Video as Always, Keep it up Pablo!
My reference XTX runs closer to 73-75c horizontal in mid tower ATX vs 70-72c vertical in a mini ITX. should I be concerned about the difference in temps? GPU is stock settings default mode.
It is normal to run differently depending on the orientation
@@AncientGameplays thanks I'll take it easy now now 😂
Very interesting topic. What about changing the PCIE cables? i would like to test that (probably a waste of time) but I will definitely try to replace my display port cable to see any temp differences
Used your discount code 2 times today man lol. Hope your getting a little kick back!
Thank you!
Hi friend! sorry I have a problem with my new Rx 6650 xt I can no longer play 4k videos on TH-cam, before I had a gtx 1050 and if it allowed me! my monitor even being 1440p gave the option to watch videos at 4k! What could be limiting it?
Do a clean install of the drivers with ddu
@@AncientGameplays Ty bro! I do it!
@@AncientGameplays thanks a lot bro it worked! in fact, thanks to you, I was encouraged to buy an RX6650 xt! Merry
new year and much blessings! :D
How can anyone report responsibly that there wasn't enough fluid, when Debaurer said himself 'I didn't measure the fluid when I cut it open'. You CANNOT say 'a few mm's too little' from a few pictures of open mesh.
Vapor Chamber requires the fluid to turn into gas, meaning it can't have too much water/fluid like a liquid cooler, it's also a low pressure container meaning the fluid vaporizes at lower temperatures, so water wouldn't need 100 Celsius to boil over, but say 80 instead?
As well, roughly 0.1% of all MBA reference cards might have this defect, because say whomever assembled it (usually a robot) ended up making a mistake.
@@dra6o0n I'm thinking it's a typical 0.1% - 1% of reference cards also.
I just done a re paste on a reference 7900xtx with the thermal right TFX and TF7 mixture, (very thick) with minimal pump out, reduced hotspot temps from 50c delta to 14C delta with a 10 hour stress test, also with thermal 4mm thick pads behind the core, 74c temp with 98C hotspot 420 watts 3018mhz core, memory is at 91C doesnt go higher than that, at UV its currently sitting at 67C 83C hotspot, 81C memory temps, success i guess (reference cooler btw)
Yes, pumpout can cause those issue. A good thermal pad fixes them too
the Shia LaBeouf "Do It " had me laughging hard🤣🤣
hehehe
Do you think a XFX 7900xt merc 310 can out perform an AMD reference 7900xtx?
Maybe it reaches it with OC yes
the problems is fixed today or still going?
these were some defective batches, you can't fix defects, the cards are send back and then the consumer gets a new one. The "Normal" cards never had such issue
@@AncientGameplays so if i buy a card today in this very moment would be a "normal" gpu? that's all i want to know.... the new batches coming to restore de stock are the fixed ones(normal)?
@@farfanSK8 yes, also this only happened on reference versions
@@farfanSK8 No. I just bought one on 3/13 and it is bad. Same issue. Talked to tech, they know nothing at AMD.
@@cfusionman1225 too bad
OP of the reddit thread just said the issue has returned after moving his PC case. I think Der Bauer has provided the reason for the hotspots along with Igor being that there is a lack of coolant.
Yes, but some users have it okay there. Very odd that only some user have this issue
The issues with the vapor chamber is only for the reference models
Yeap
Hi again Fábio. Looks like my reference XTX reaches the 110C with the power limit +15%, while stock it stays at about 90C. Do you consider this one as a faulty card?
if it is 90C at 350W seems fine to me as the case airflow also matters a lot. With some tweaking you can decrease them even further
can you recommend a hdmi 2.1 cable manufacturer should i go with club 3d too?
thanks for the tips, i m actually in this situation (7900xtx) with some hotspot at 110° and higher, i will try to switch my HDMI cable to a DP port i guess and see if it resolve the issue
Hope this helps
My 7800xt runs around 58c but hotspot is around 90 it was closer too 100 but i adjusted fan curve
that's a HUGE difference
Your intro's make me laugh. I need it after a bad day. Thanks for the video. AMD needs to fix this and save some face. I hope.
Thanks for watching and commenting!
What worked best for me Fabio was cleaning out the thermal paste and then using a Thermal Grizzly Kryosheet, this stuff is magic! I saw many people referring to the PTM7950 however I could not access this in my country, I managed to get the Kryosheet and my Hotspot and GPU delta dropped from around 50 degrees celsius to generally under 20 degrees celsius. Only down side in my opinion is that it is electrically conductive so you have to be careful when putting it on the die. I figured my problem on the 7900XT was due to pump out as thermal paste only improved temps for about a week before they got worse again. If PTM7950 is availabe it might be a better option as it does not conduct electricity however both are great products!
Glad to see it helped!
What if there are two problems? Try those weird cables on an nVidia card...what happens then?
They definitely are, that's the point
Drama is fun when I'm not involved
pahaha
I was getting bad temps, crashing with 3 cables rated for 8K/hdmi 2.1 and wasn’t gaming in but 4k or 1440p. I also had a very thin, but from top to bottom pixelated verticals line.
I made sure bios Was up to date. I removed Nvidia drivers. Factory reset the 7900 XTX MERC10 4x and my fix came with a Nintendo Switch HDMI. Which pulls 120hz. Gpu usage and temps came down dramatically. I can even play with overclock settings and got 17599 on Port royal. Factory cooling.
To tell ya how bad things got, it was like I had to treat this like a 1080P card coming from an Rtx 3080. I still have a 5900x.
I also have a gaming x trio 7900 XTX. Which has had zero issues. Merc10 hasn’t had issues for awhile since it went away.
Another possibility is that the games in combination have been messing with the GPU. Like I was playing Fortnite and was messing around in settings on 1440p, getting a feel for the ray tracing in addition to handling the most basics on epic with tsr epic. I still had around 58-62% gpu utilization. I believe my run through the situation, includes playing Fortnite with ray tracing on while having the Radeon overlays on to view temps. Nothing seemed out of whack until it was. Zero over clocks performed, or memory, undergoing or any performance tuning for that matter existed before the issue.
Since we’ve had a Fortnite update. I still play with lumens and hardware ray tracing off.
I’m not surprised that Fortnite can be demanding. I remember when Rtx 2070 super handled 1440P, then Rtx 3080 handled 1440P dx11 Fortnite. Now even with 4080 or 7900 XTX, the quality of the game can be so high that 4K is not well at native unless the rest of the settings are low.
It looks to me like one or more batches have no (t enough) fluid in the vapour chamber.
This explains why it works vertically because the little amount of fluid there is has a chance to drop back down.
On yours it doesn't matter because it has enough fluid.
That's what I think anyway. When your sweet cooling solution fails because you forgot a drop of water. 😂
Hey fabio, should i use avx with prime95 small ffts i have no clue.
Avx is heavier, so yes
@@AncientGameplays Nice okay thanks🤠👍
Hey my card gets 70 to 73 and hotspot goes up to 90 - 95 is this normal????
quite normal yes, especially with lower end models
@ I have the Asrock 7900xtx Phantom Gaming OC and when I play helldivers on 2k or 1080p with supersampling it actually reaches 90c hotspot.
Wow, this is really surprising. I'm working on SPS power chips, not on GPUs themselves. And in my limited knowledge there isn't a way that the IO part of the GPU should influence the core temperatures, but seeing as it's not an isolated case. I'll keep my eyes open to further developments
So the thing is, it's not the core temp that's high, it's junction temp. Junction sensors are a network of sensors embedded in the die, the AMD software is either displaying one of many or one particular one, so it's actually possible that the one being displayed, or the hottest one is localized in one part of the chip that's doing one specific thing.
@@PineyJustice In my head I imagined IO to not be part of the die and was a separated chip, but thinking about it and considering the huge bandwidths we're having these days I guess it'd make sense to actually be on the die itself. Still quite an odd behavior.
It'd be interesting to see if connecting that Dell cable (or any with pin 20 wired...) to an Nvidia card or a previous Gen AMD would result in any change in temperature.
Weird fix, but I can only imagine the improvements of this fix + vertical mount, or a custom card that has a cooler that doesn't have this issue.
Wow. I was thinking i was gonna have to repaste my gpu because my hotspots almost stayed at 110º. Swapped my dell alienware 4k monitor from a dp to a hdmi and my hotspot went from pegged at 110 down to 80-90 during gaming.
Great to help!
What software did you use to overlay the stats?
I hope der8auer transplants a known-good vapor chamber (like yours) to one of his cards that is overheating to see if that fixes the issue. If the issue is resolved after swapping the vapor chamber, then we know it is indeed the chamber, but if the issue persists then it may be something else.
You would not be able to tell by doing that though. As one person sent him a card that was thermal throttling. A change of thermal paste fixed the card. So when you swap the coolers. Thermal paste would be changed. Would it be the cooler or the thermal paste ???
@@suzie9874 He already excluded the thermal paste, though with a very small sample size of course. So I think it is less likely to be thermal paste and more likely the cooler or some other weird issue (as discussed in this video), or more likely a combination of both as the temperature changes with different orientation and different cables.
@@myname7021 I just seen that they also have an issue with higher heat with 20 pin monitor cables. When using 19 pin cables the temp would lower. Odd i know but i did see this somewhere.
@@suzie9874 That is very weird. I hope AMD get's to the bottom of this ASAP
Maybe that's because it is connected to the display engine in the die chiplet and mess up the temp..
2:36 " I just cut my cable to check, it only has 19 pins"
comment below - "You could have just tested the pins on both ends with a voltmeter..."
:'DDD
🤣🤣
Will undervolt the GPU (main frequency 2400mhz and Max 2500mhz & voltage 1120Mv) can reduce junction temperature??
it will, a lot
I finally got some advice about my new 6700xt (used from a crypto miner). The person apparently had had the same exact problems. (bios constantly resetting and not restarting from bios without hitting the reset button). He seems to have thought I got a bad card so I had 20 days left of warranty and sent it back. They're going to send me a new card free of charge so I'm very happy with the outcome.
Thank you so much as always for the info Ancient
My pleasure!
De momento ,los que intentan tramitar su grafica con el problema de sobrecalentamiento por RMA , no son aceptados ya que dicen que esas temperaturas son normales y no les tramitan la devolucion ni les dan una solucion .
Now that's bad pr and should not be allowed. Gladly stores have to take them back lol
Wonder if Fiber Optic HDMI cables suffer same results 🤔
who knows I guess...
Atomized Android and I told you, on your Discord, what the problem was and how to fix it.
So let me repeat it here again.
The working hypothesis is that it has to do with the very narrow fin pitch of the GPU cooler (the distance between the heat exchanging fins).
This has been tested by Atomized Android a number of times on his 7900XTX and is repeatable, and during testing his GPU reaches 400 Watts TBP.
When running the card horizontally, the narrowness of the fin pitch does not allow for the air to be ejected out of the with enough force (fans don't have enough Static Pressure?) and so the warm air is being sucked into the fans again and recycled.
Atomized Android with his 7900XTX reference design card came across this by accident.
He was using is previous 6900XT with a program called "Fan Control" which would ramp up the front intake fans of his case when both the CPU or the GPU came under load (unlike normally where you have to choose if either the CPU or the GPU controls the fan speed of the front intake case fans, but not both).
When he put in his new 7900XTX the profile was no longer applied, because it didn't recognize the GPU (it was configured for a 6900XT).
When he was running his tests, the GPU went up to about 108 degrees Celsius and he was a bit annoyed.
He noticed however that the front fans in his case weren't ramping up the way they had before, although the fans on his GPU were going like the clappers.
So he looked at his Fan Control profile, noticed the problem, configured it the way it was before for his 6900XT, ran the test again (this time with the front fans ramping up to blow air over the card when the GPU came under load) and his max temp at a 400 Watt TBP was at around 84 degrees and the delta between the "Current Temperature" (edge temperature) and the Junction Temperature was around 16 degrees.
You showed Der8auer's efforts at getting to the cause of the problem and the only bloody thing he didn't do was push some airflow over the card with a fan when it was horizontal.
This would explain why the problem is so random.
It would be bloody easy for you to test this with your system. All you would have to do is have the intake fans of your case set to various speeds and see the difference it makes to the max temp your card reaches. Start with full speed and then ramp down until you find the point where your temp is optimal.
Remember if you have your front case fans set to ramp up dependent on CPU load that games don't really stress the CPU and the better the cooling you have on your CPU the less likely your front case fans are to spinning up.
So there you go, job done.
I got it, and will report it as well (and thanks for doing it) but these seem to be different issues here
Wazaa mate, great video.I watched other videos people with out any issue.
same here
My question is: has anyone found out what or where about the Junction Temperature?
What do you mean
@@AncientGameplays I mean if anyone knows where dose the place that the 110C temperature take place? It doesn't look like the core or ram
Is 95 degrees ok hotspot temp to for 7800xt
hotspot? yes
You can undervolt though
I brought a new cable i feel like its dp cable 1.4 from cable creation now my hotspot temps reach upto 100 degrees in an instant but it happens mostly in rdr2 for some reason
The Vapor Chamber is the problem not the cable,
The issue is the 20 Pin cables are causes the port and controller to heat up, which interferes with the Vapor Chamber's wicking and return paths for the condensation.
Putting the card horizontal causes heat to pool on the controller.
One doesn't invalidate the other
@@AncientGameplays Likely cause the cheap cable doesnt enables Sync Options, allows the GPU to run full tilt in gaming, where as the better cable syncs the GPU, and limits the rendering.
Either way, There's cases and results that prove an Display Port cable isnt the problem.
@@SkateZillaSimulations nothing like that, in both cases the power draw was the same
@@AncientGameplays Power Draw fluctuates during rendering, these could be cherry picked screen grabs, even the order of the layout changes, last time I checked, changing a cable does not change my OSD Order and Layout.
Pin 20 Problem was fixed 5 yrs ago by both AMD and nVidia Thru Driver updates.
TO Prove a Cable would do this, there needs to be a Video, of the screen and card, showing when the cable is switched while running the bench mark, the temps drop.
Shutting off the benchmark to switch cables allows the VC to cool and restore functionality.
AMD Already announced they discovered Vapor Chamber defects where the VC would "Vapor Lock" and the condensed fluid would not return to the GPU plate to continue cooling.
So the point is moot.
The display port on this generation is not the same, that alone should demand a next gen, certified cable.
Also, the vapor chamber is something prone to all sorts of problems, why bother at all??
Not if well done!
AINTNOWAY, AMD was mocking nvidia and now this. They are so inline with eachother that even mistakes are relatively the same.
not even close to same lol
Yes not even close to same. Nvidias one was user error. Amd straight up released a defective inefficient gpu
@@anuzahyder7185 so, 2% units with a defective COOLER, the bias is real there. Look bad at nvidia's history then
@@AncientGameplays bro give it a rest.
A wild guess for me ,maybe the dp 2.1 send more power to the cable so the cheap cable cannot handle gussing problems, but that is just a maybe......
A podsibility yes
Thanks I decided to try and buy a 7900 xt merc (the highest boost model) for 900$ on newegg. I bought it but then later it said i needed to have an adult signature when it comes. Why would newegg take my money before they tell me that? Now i haft to wait 3-5 days to get my money back and also 4 months so i dont haft to go through that bullshit again.
Lol, newegg
@@AncientGameplays Thanks to the fraud department i got my money back earlier lol
I ran furmark for quite a while last night, didn't go over 81c, so fingers crossed for my unit. 7900xtx powercolor MBA card
oh on a standard horizontal fitting too
tell a lie, with your max overclock settings junction temp hit 109c on furmark in about 2 mins, so turned off the benchmark, but guessing this is extreme and under normal settings ran it and it seemed fine
Same here but Sapphire
Cool man, yeah so all seems good, I did get a little panic on when I used your max oc settings, and then ran it again, shot up to the 110c before I switched off the benchmark, the temps really shot down quickly to normal which I was happy with. I ramped up the fans to 100% and with oc temps still shot up, took about a minute and a half to two mins to reach 110c but literally second to drop back down when turning off the benchmark. I since read in the furmark notes say not too stress test with it on oc settings, whoops! :-) but stock worked ok 😅 and all looks good for me, hoping not to many people suffer with the problem. Loving the card so far 😁
Hey is evertyng still okay with you card 7900xtx no problems with heat i want to buy it but im litle scary about temp?
Some YT channels pointed out problems with the vapor chamber (not enough water? gravity bending the chamber?)
That is yet to be proven as not all models have those issues. May be for some though
Der8auer really needs to open up a vapor chamber on a known working card to know for sure what is going on. But people aren't in the habit of sending away their working products to be destructively dissected....
In his video he talks about how he sent some 4080s & money if they did not want that as well if not one also said if you can send me a AIB model some would take that as well. So any card that was sent he paid for or replaced it for what they asked for. I think He got 3 shipped in if I remember right. I get some might be afraid but Der8auer never risk his name & not give the card back or something in return.
Nice video like always - for cable try whit your 7900xtx The PremiumTech Europe DisplayPort 2.0 Cable and whit last video of yours overcloking and undervolting 7000 cards temps shall be ok.
I think they meant that the 20 pin is better than the 19 pin because one of the 19 pin in the 19 pin cable is NOT grounded!
Maybe
I got my 7900 xt merc black edition :), lowered voltage to 975mv instead of 1100mv i think im around my limit, i wish i could edit power play tables with mpt I would get way way higher clocks.
AMD did not think this one through
They did. They don't make the coolers though.
Has anyone noticed Asus newest bios x570 the 7900xtx running 8x 4 pcie lanes but still runs exactly the same ??? 462w 100% usage runs flawless
Fabio Pisco, you slapped my hear and it made like a butterfly brah, guess I will have to buy another one or use Frame Generation, GG.
DP cables gonna be getting some shady marketing. What if those comments were made up so that cable sellers start some expensive sells ? 🤔
Not the case, I've known for YEARS that a good cable should always be a primary thing because it could cause issues like flickering, blackscreens and even booting problem. Its a known thing.
like this though, I never saw, but since the new AMD cards use chiplet design and a new Display Engine, it may be possible
WTF this absolutely worked for me. I had a dell displayport and swapped it to a different one and went from 67/108 hot spot to 50s/low 90s. Im really curious what the actual issue here with this.
edit - My card is an Asrock 7900xtx Taichi white. This fixed my heat in game but my card still sucks. It goes unstable the moment I undervolt even a little in games but I can undervolt it hard for benchmarks..rip
That's due to power scales and the games you play. Because benchmarks are usually heavier, meaning that the "base" voltage is higher since it is using a higher power state.
In games, some can be use a higher base voltage and others a lower one, meaning that with the same slider option the voltages will be different. I myself only use down to 1130mv. Want to really go power saving? Decrease the frequency
@@AncientGameplays i have to sit at 1150 but interesting. I wish I knew this before spending a night running 3Dmark getting higher and higher scores while overclocking to have my night shattered by the first game lol.