Explaining junction temp Because people Dont listen!!! 110 Degrees is not ok!!!
Вставка
- Опубліковано 5 жов 2024
- Explain junction temp Because people Dont listen!!! 110 Degrees is not ok!!!
/ notanapplefan1
discord: / discord
patron: / naaf
paypal:paypal.me/nota....
AMD didn't say 110c is "fine", that's just the click-baity headline people use, this is the actual quote for anyone who cares; "Instead of setting a conservative, ‘worst case’ throttling temperature for the entire die, the Radeon RX 5700 series GPUs will continue to opportunistically and aggressively ramp clocks until any one of the many available sensors hits the ‘hotspot’ or ‘Junction’ temperature of 110 degrees Celsius. *Operating at up to 110C Junction Temperature during typical gaming usage is expected and within spec* "
Yes and that statement is very poor, you know that right? Would I get infinite performance if I chill the card? Nope.
It doesnt ramp clocks till they hit 110c. Cause if they did and you keep it infinitely cool, it would overclock infinitely. That's where they confused people. Plus 110c might be within spec, but it shouldn't reach that temp during typical gaming anyway
By the way' is expected and within spec' is pretty much the same as 'is fine' in my book. Lol.
They can't stand that $ for $ the new entry level 5700XT stomps the 2060Super.
Expected = Normal.
I'd be funny to see how these same people would react if Nvidia showed junction temp, I can only imagine suddenly it wouldn't be a problem!
Same goes for ray tracing. As soon as AMD will have it, you, fanboys, will praise it
@@noctrk161 Yup. Just like we praise freesync, CAS and Anti-lag. We praise AMD coz they do it better and not at a high-cost, performance impacting shit like DLSS, RTX and Gsync
@@noctrk161 don't get people wrong who condemn RTX now. Speaking for me and for many more ppl, we know Raytracing is the future. But its not quite there yet. Nvidia should not have pushed RTX like they did at that time. You can even clearly see in Benchmarks that when doing RTX operations, the Game Performance drops significantly. But you see that it is due to the RT cores being not strong enough (there are not as many as needed).
When playing Games with RTX on, the framerate tanks, but the CUDA cores are not even close to maxing out, as they are limited by the slower rt cores.
This is the fundamental issue with Nvidia RTX. The 2080ti should have twice the RT cores and every other card should at least have as many RT cores as the 2080ti does now.
In a few years Raytracing will be there and it will be great, it will be glorious. But not today, and certainly, as we saw not in 2018 either.
I have Nvidia card and I will praise raytrasing when it does actually look good and run at good speed! Not untill that happens and at this moment most raytrasing stuff if bull... and speed is... well bad...
But I am sure that it will get better by time. At 2023 we have more speed and programmer have maybe learned to do raytrasing so that it actually improve the experience. But that is still few years from now.
@@noctrk161
Lol my problem with raytracing is that not enough games have it, which is why I haven't picked up an RTX card.
Even if it came to AMD, it still wouldn't be useful if there's barely anything to use it with.
I bought a RX 6700 XT a few weeks ago. It is a good example that you have to have a case and cooling to go along with it. I put the card in my mid-tower case. I started to run games and the GPU was hitting 80c, but junction temps were toward 110c. I am an engineer, an worked on electronic products for 18 years. A lot of components were rated at 104c for highest temp before problems. So I looked at the case. I added one fan. Put all fans directly to the power supply molex plugs. The mainboard fan connectors were to slow for cooling the case. So I have 3 fans air intake, and two fans air exhaust (including power supply). There are grills in the case the intake air can also escape by pressure differential. I set the Graphics card fan profile to on early at 60c to get ahead of GPU, and Junction temps early. The card fans hit a steady 2600 - 2900 rpm and keep the junction temperature down between 80c - 90c. The GPU runs about 65c, and the CPU 70 - 75 c. I happy with this level, and I don't smell heat, like to hot anywhere in the case. At 1440p this card performs well, about 120 - 200 fps depending on the Battlefield games I am playing.
Nice!
@@dane4890 I have since done some more experiments. I capped my framerate FPS in Adrenaline at 100 FPS and that really cooled down the card. I see no difference at running 100 FPS vs 150 - 200 FPS on my 1440p monitor. I run monitor at 1080p which still comes out sharper on s 1440p monitor. Don't need 4K. I have my 3 fans now on my AsRock Phantom Gaming 4 motherboard and set fan speed profiles in BIOS according to processor heat. Still cap 100 FPS and CPU runs in 60 c range, graphics processor 50 - 65c and junction temps 60 - 70 c. I run a gaming fan profile also in AMD Adrenalin.
I have the same problem MSI MECH RX 6700 XT. Full GPU load temperature hot spot are 105~
@@skitzdaniel3314 I set mine at 100 fps also, running great
taking notes from this hahah. same card here and was playing bf1 earlier on near max settings capped at 144fps and was hitting 90s on junction temps and around 80-90 regular temps. getting a used new case tomorrow with some noctua fans in it as well as being able to add my old ones so hopefully overall temps will go down! just need to learn how to do some fan curves lmao
The junction temperature is the temperature of the transistor junction. The case temperature is the temperature outside the die, on the heat spreader. There is some thermal resistance in the heat spreader, so there is some temperature drop across that resistance (analogous to voltage drop across an electrical resistor). Simple as that. Every IC package ever manufactured has a higher junction temperature than case temperature, it is the nature of reality.
It's worth mentioning that the ability to read junction temperature means that higher edge temperatures are ok, because we no longer have to be as conservative. In past hardware, we maintained a lower temperature envelope because, without knowing the hot spot, we had no idea how far we were pushing it. This became particularly important in instances such as older CPU coolers, where the heat pipes didn't make up the entirety of the base, so the chip had large potential to be unevenly cooled. We would stick to CPU temps under 80 not because 80 was bad...but because we didn't know the hotspot and didn't want to risk it. If, say, the chips lifespan became unusably short after 110c, we would want to stay as far away from that as possible, to know we aren't hitting it. Now we have a way of knowing, so we can push those limits more. And the stock temperatures honestly seem a bit conservative anyway. Radeon VII allowed 120c junction while overclocked, which while I can't imagine is good for the chip, wasn't considered terrible enough that AMD wouldn't allow it to run. 10c on the top end of thermals makes a huge difference...chip degredation isn't linear. That less than 10% difference in temperature will make a much greater than 10% difference in chip life. So assuming we can read the junction accurately, 110c seems like a conservative temp for 7nm. I don't think this cooler is good. It is louder and hotter than aftermarket coolers. That being said, I don't think it is unsafe for the chip, and I think the chip running hot, even at edge temps, is less "we're pushing the chip too hot" and more "now we know exactly hot hot it runs, and how far we CAN push it, without being excessively conservative." Revealing the junction didn't make these chips any hotter. But it did make the drivers more aware of what the chip temperature actually is, and assuming we know what the safe temperature for a balance of performance and longevity is...being able to throttle less conservatively is a good thing. The cooler sucks. But it's no worse than any other blower. The chip itself, though, is smarter about it. Even if the reviewers do point out the edge temps...they no longer matter. Because those were never the dangerous temps, those were a range we used to assure we wouldn't hit dangerous temps when we were unaware of how hot the chip truly is. Now we are aware, and the edge temps are essentially a product of the past.
not to mention AMD has said 110c is nothing to worry about and they can be pushed past 110c so all this worry about 110c is unfounded
@@adammartin4906source to this?
If Nividia showed their junction temps (which they never will), then I suspect all of this nonsense would suddenly go away.
they show it now and its high as hell
have a 4090 that has a GPU temp of 34c while gaming. What do you think my junction temp is?
@@davehenderson6896 what are you playing? A 20 year old game?
No all modern day AAA games.@@Dell-ol6hb
Thank you for this video. Being a new person to PC, having conflicting information about these temps really had me believing that my GPU was broken. I was seeing 85-90 temps being reported, and seeing online that a GPU should never run that high, not realizing that I was seeing JUNCTION temps, had me confused and defeated esp in this GPU market. This cleared it up. I really have 70ish gpu temps when under heavy load. I really appreciate it
so bro my gpu is rx5700xt and in warzone2 my gpu temp get 75 it is ok?
@@Marwan_Realtor 75 celsius would be great.
Does 110 junction temps is good?
Bro my gpu is Rx 5600xt and in warzone my gpu temp is around 90 to 95 it is ok or bad ?
@syedabdulsami2370 average or hotspot? That's pretty high for hotspot but too high for an average temp. I'd be curious to know what your hotspot temp is..
More information is a good thing, disinformation is a bad thing.
Limiting fps to 60 and using fancurve helps a ton, ran 85-92°c junc temp before with stock settings, after fancurve junc temp max 62°c. Huge drop...
with AMD, limiting FPS is the key.
I think AMD was right to show us the Junction temp, but i also think the situation is like the situation we have with modern cars. Nvidia is like a off the showroom floor Corvette, its fast and you can put some bolt on upgrades on it but the inner-workings are mostly hidden, like the temp gauge, its not really a gauge its an idiot light that looks like a gauge, it only really had 3 positions, Cold, Normal, Overheated, the small fluctuations within the normal range are hidden. AMD is like a Lotus, or a Tuned RX7, good off the show room floor, but a tuners dream, and all the inner-workings are exposed for you ti see, you get real working gauges that would scare the shit out of a normal person buying a corvette. Most people choose the corvette... and AMD is like the tweekers dream. Most people arn't smart-enough or don't care enough for the advantages of having an AMD card to matter.
The power draw on your GPU is going down because it's running at a lower temperature.
The electrical resistance of undoped silicon decreases with temperature.
As a consequence you have more electrons 'leaking' at higher temperatures; The electrons dissipate through the undoped silicon without doing any useful work.
These useless electrons sill have to be provided by the VRMs though, resulting in a higher power draw.
If I remember correctly the difference is about 4% per 10 °C.
i got a ryzen 2700 for 140 euro really happy
nice thats a great deal you get their
@@hardcoreteunes i had a 1200 sold that and bought the 7
2600 here for 75 EUR. It was a private sale but new, boxed, unopened. Fantastic performance for the money.
@@tnetroP dude thats insane
@@tnetroP Amazing deal.
Asus is in full defense mode about this 110 C press release.
I wrote about my GPU VRM being 110 degrees and got a reply that 110 is ok for GPU junction temp, together with the AMD acticle about 110 is ok for junction.
Had to write back that my Junction max was 88, so that was not the problem.
Lars Jørgensen looks like they are trying to turn away any unnecessary RMAs
Vega 56 Strix owner here.
My VRMs got up to 117°C. Contacted Asus .. they said, that the VRMs are rated for 125°C and anything below is completely fine.
(The card turns itself off at 125°C VRMs)
Yeah, so for Asus there is no reason for repairing or taking back the shit product they sold me. Thanks, now I kown which brand I will never buy again.
Finn bro that’s fucked man I feel for you wth . I just bought a ready built pc a few days ago running asus 5700 x I’m getting paranoid at the temps
Everything I've seen lately puts AMD chip temperature in line with Nvidia's. The duel, and triple fan models lower the temperature 10+ degrees. The Red Devil is supposed to be very quiet.
The best way to think of a junction temp is like when cooking a chook in the oven. You might have a high surface temp (junction temp) but the temperature of the meat is what matters to cook it (core temp). If you try to cook it too fast, you'll burn the skin without cooking the meat, so you have to lower the overall temp and increase the cook time (thermal throttling). Luckily, with GPUs this happens automatically based on the junction temp. If only ovens could do the same.
you have it backwards. The junction temp is the meat in this case cause it's inside the chip. And that is hotter than the die itself
Does anybody else think the title of the video completely contrasts with the content of the video? The title, where it says "110°C is not OK" makes it seem like he's going to say AMD screwed up, but then in the video it seems as though he's expressing that the temp is nothing to worry about.... Might want to be more explicit in your titles man.
Left me hella confused
Just got myself a 6600xt,coming across this video put me more at ease,wasn't really too worried about my hotspot cause i heard it was fine,the highest the hotspot has gotten on my card is 98c and that is with really intensive loads like Furmark,under normal use i never get that high,i sometimes get into the 90s with really intensive games only because i pushed my power slider to the max,the card never goes above 71c under max load so im happy with the card.
This video is missing the Sith robe
I would actually trust Engineers with degrees in electrical engineering then click-baiting rant boys.
"click-baiting rant boys" - I agree, thank you!
I'm confused as to why AMD tell us that 110 Degs Jcn is Ok (for my RX6900XT) and the rest at around 100C; which I thought was bad, but they say is normal. But others say is not. Who really knows with modern tech. Companies lie, Governments lie, podcasters lie. Mysteries. Still works a year later.
It's not a real 15 degrees outside, it's an Irish 15, so more like a "feels like" of 10. It's miserable windy and wet. :(
Here in Floripa Brazil, its the same 15c that feels like 9c
living on a big peninsula
Messing with your gpu fan curve helps with the junction temps, if it bothers any of you. My Junction temps ran around 80-90°, until I messed with my gpu fans in the AMD performance settings 👌
Thanks for this, I've been searching up and down for whether I should be worried about this and all I see everywhere is 'it's fine at 110' (I am not a massive whizz!) or 'IT'S RUNNING HOT OMG'. Simple explanation is much appreciated. Cheers mate.
Didn’t we know this junction temp thing when vega came out? I know I did. Maybe the ram is locked down as AMD May have faster cards coming and want all the line up to have some space between them?
I was just wondering what it meant. Under heavy Load, is 75c Junction Temperature fine??
THANKS FOR THE VID! my 6750xt Powercolor Red Devil has 90C junction temps on Diablo 4. thanks for putting the Red Devil on the image helped direct me here. what i noticed was the Red Devil doesn't kick in fan speeds until late in terms of heat distribution (unless you manually set the fan speed). i think thats the reason for 90-95 junction temps on it even with such a massive cooler (for its class). you should definitely make video's like this more often i like how you describe things very straight up.
The reason your card is in the 50's is that the memory doesn't clock down at idle on high hz display's (bug).
If at 144hz set monitor at 120hz and see temps drop to 30 - 35c when idle
What if i am playing on 60hz tv? seems worse
My 5700xt at 110 junction runs at 1970Mhz almost constant without thermal throttle. (Undervolted) However I was able to bring down junction to 92-94 running with stable clock speeds around 1920-1940Mhz. The result is incredible - same performance and more stable experience.
What did you do bro?
@@ralphmatthewmiranda9531 undervolt + underclock probably
@@mertince2310 i have a 6950xt and my junction was hitting sometimes even above 110 degrees so I decided to undervolt and underclock by around 100 MHz reduced temps of junction to around 70-80. I probably will reduce my fan speed now tbh
while im playing wow my RX 6700 (non-xt) Sapphire Nitro+ hits around 70-75c and junc temp to 90-95, i am worried about it. My case isnt the best so would you recommend me to upgrade my case to reduce the temp a little bit?
@@-MertArda thats fine
Junction temp is memory temp? wtf are you talking about?
I bought the 51risc RX6600M from AliExpress a year ago. I changed the thermal paste because it was dry. I used the GD900-1, it didn't help at all. The GPU junction hits 105ºC, the GPU edge 73ºC and the GPU mem 70ºC.
So I'm guessing junction temps don't really matter as long as the cables are not burning themselves or the GPU isn't setting it's self fire it's most likely fine. So basically people are panicking over nothing
nope sorry,
a normal junction/hotspot temp should be 80-90c at most. These cards will die alot faster because running like 80-90c (temp of gpu not hotspot/junction, that would be more like 110). Basically the paste and thermal pads will run out way faster from the constant heat. Also ram chips should be 60-65c at MAX no more because it will kill them so much faster. Basically if you want the GPU to last more then a year or two gaming on it and not have dead GPU Ram Chips or having to replace the thermal cooling constantly(paste and pads alot), then watch your TEMPERATURE pleeease. If you keep your temps at a reasonable level and not too much then it will last longer.
Basically: heat (80c-90c+ all the time) kill computer parts faster and degrades faster
About cables: no cable should be getting that hot unless the cable is broken and shorting(touching something it shouldn’t releasing electricity or pushing too much wattage or horrible 3rd party cables that short like stated before) like 4080-90 cables.
Anyways a cable should be the least of your worrys
This is a reason why I havent upgraded from a 1080ti because even the 4000 or 3000 series run stupid hot
Mate you're losing performance due to thermal throttling. Imagine you're only getting 80% of the actual performance while paying over $1000 for the product.
I'd love to see NGreedia's Junction Temps, I'd be willing to bet they're right on par with AMD's if not higher considering their Clock Speeds...
Exactly. It would be very foolish of Nvidia to set the boost frequency well below the safe maximum operating frequency specified by TSMC and leave lost of performance on the table.
Nowdays when we talk about amd gpu we talk about underclocking and undervolting.
5700 xt i got is set to run at 2100 Mhz with 1.150 V reaching over 80 C in gaming. 50 C in idle. Power consumption over 220 W.
Underclocking to 1950 Mhz with 1 V power consumption downed to 160-180W with 41 C idle temp, while in gaming edge temp goes to 70 C and hot spot to 85 C max. Fans are set to 60% max. So huge difference from default settings. I tried FurMark once for a minute and hot spot easily reached 106C while edge temp was 79 C. so the gap was almost 30 C difference. I dont surrges any sintetic benchmark since it gives artificial behavior on chips that you can not meet in casual daily activities.
Anyway, 5700 xt is one hell of a chip for around 350 - 400 euros.
I've recently set fan speed curve to 25 degrees - 50% with mild increase to like 65 degrees - 75% and so on. It can get kinda noisy but it keeps temperature at around 70 degrees max. I had problems where my pc restarts for some reason and I've figured out that it's probably because of temperature, which was getting to 80 degrees and running like that for hours on default fan settings. Anyways, it's amazing card for 350-400 euros looking at it's performance, a little bit noisy but nothing that headset can't solve.
So what you're saying is, if it isn't hitting 110 consistently at junction temp...and the edge temps are staying sub 80....i'm good? no stress? everything fine?
This video has aged well
Agree with the junction temps and the death of overclocking. But not so much on the memory argument. Infinity cache must be doing something right because if I OC the memory I see absolutely no gains, maybe a few points here or there, but that can just as easily be a result of having a colder run (longer time between tests).
Overall the 6950 XT is a weird card. I dropped on a standard Sapphire Nitro+, even though I had a Sapphire Nitro+ 6900 XT. Mostly because I knew the 6950 were essentially higher bin GPUs and could potentially go further on OC. Which mine can, ever so slightly. But overall my 6900 XT Sapphire Nitro+ can do a rock solid stable 2685MHZ at 1080mv (will even do at 1060mv, but I keep it higher to insure stability), and a slightly less stable 2710MHZ at the same voltage. My 6950 XT will not go below 1120mv at any speed lol. Not even the default clock. I’d probably have to drop to 2300MHZ or 2400MHZ. But the OC headroom, should I leave voltages default? Get to about 2800MHZ. So the card was definitely designed to be on water because even if I leave the card on Rage Mode? The junction hits 115c in benchmarks because the stupid Rage Mode profile feels it doesn’t need fans (must go by the overall GPU core temp). If I just roll with the standard onboard OC BIOS it sits around 95-105c junction. But if I setup my own profile and max the fans out and leave all the clocks standard OC BIOS speeds and voltages? It see between 91- 101c. 1.2v is just too much on air. And while I can drop to 1150mv it’s just still too damn hot.
I just don’t trust factory clocks because everyone is so concerned about silence that AMD and AIB partners are forced to error towards heat output.
110° junction temps problem is about a limited boards, my 7900 xtx MBA works perfectly fine max 92° at +15% power draw horizontally
I heard the 5700xt runs at 110c + :) lol Love you paul
"If you want to talk about stuff, get your FACKS straight!"
It's not that bad GDDR6 memory chips are used. It's a limitation of the memory controller of Navi, you can see even 3rd party cards not getting higher memory speeds. Even using Samsung or Micron memory does not matter.
exactly if gddr6 memory couldn't be oc'd then you wouldn't be able to do it on rtx cards as well. (haven't used it but I assume you can raise clocks on them)
Just here to say after 4 years my MSI RX 5700xt gaming is still alive and kicking hitting 110c junction temp is no sweat. Gpu temp constant between 60-80c.
Still fine?
New to amd cards and researching the different sensors reporting temps. I understand junction temp is like a Hotspot temp and 110degrees is said to be normal but I don't feel like it should be ignored and it seems with proper cooking applied no part of the gpu die should be sustaining anything near 110 degrees.
Hey man i wanted to say i dropped my. Navi temps down on full gaming load to arounf 67c i changed the thermal.pads to 0.5mm and put grizzly paste and washer mod. Amazing results man
Power consumption goes down when it is better cooled. Even if frequency goes up colder silicon can potentially be more power frugal.
Yeah I have seen too many idiots using junction temp instead of core temp. And the problem is that some are using it to make AMD look bad.
Just built my first PC and was trying to find something explaining what the heck junction temperature was. This eases my mind.
2:12 min are I'm the only one who thought that my Windows 10 got a notification?
If it's "boosting" up to that temp, that doesn't mean it's throttling. This is one of MANY factors, that if it hits that its the limit. There are several other limits
I just got my 5700 reference card and it's nowhere near 110C junction temp when testing at stock, it's like 95 peak. That's the worst cooler available and it works fine - yeah that's hot, but still 15C below what AMD spec'd the card to safely operate at. Undervolting down to 950mv and playing a little with the fan curves in wattman drops peak junction temp to 86C with very little loss in performance (about 1.5fps average in Heaven 1440p ultra bench).
I've lately been wondering if they got to Steve as well. He is nothing but convincing people not to buy AMD.
Johnny Steyn - Real Animation which Steve are we talking about here? GN Steve or HUB Steve
@@tireshredder4208 I assume GN Steve. He was making a massive deal out of the Tj Max thing...
I suspect he's received a very generous donation from certain parties.
Yep he is a real nvidia shill
...he recommended them as price to performance kings in chip department.
...he recommended the sapphire pulse 5700xt in the graphics card department.
Where in the fuck is he not convincing people to buy AMD?
There's a big difference between boosting UP to that vs throttling. Calling it throttling because it won't boost beyond pushing that is a misnomer. It's misleading.
Hotter silicon draws more power due to leakage.
Plus, increased lattice vibration (due to higher temperatures) hinders the current flow - if I recall my A-Level Physics correctly.
Just played Modern warfare 2 campaign on the red devil 6950xt. I was freaked out by the game throwing pop ups on the screen that my gpu was overheating. I saw the core temp and junction temperature in the adrenaline software and realized the game was reading junction and not the core and that was why I was getting the warnings. Haven’t really seen anyone else explain that playing with the junction temperature at 90 degrees C was ok until this video. Good job and thanks
Exact reason why I’m here, mw2 is saying my gpu is running at 92c and was showing red.
dude. undervolt it
Nicolas11x12 English
originally reported the wrong temps using the Junction temp in his RX 5700 XT review and took that video down and made 2 videos apologizing and explaining his mistake. Gamers Nexus and Hardware Unboxed, LTT and Jaystwocents all explain the GPU temps correctly. Let's be honest AMD reference models run hot and the AIB cards are what everyone recommends being cooler and quieter. AMD has made some errors with their marketing statements referencing 110 degrees being acceptable as everyone knows heat kills electronics. They should have just not said anything and really should have put better coolers on their reference designs just like they did with the Radeon VII. Heck even their CPU marketing messed up with the boost clock marketing which causes a small controversy when there shouldnt be one as their CPU's are really great. AMD just needs to be smarter with their statements so they don't cause issues when they have very good products.
Nvidias will be just as hot if they would ever show it, but they of course would not. I am glad I went with AMD, got the 6800 XT Red Devil Limited Edition and I fucking love it.
Hotspot temp is the only temp sensor we should look at really
I have a 4090 that has a GPU temp of 34c while gaming. My fans don't even turn on so no problem with coil whine.
The memory won't overclock because it's too hot with that cooler. The Sapphire Pulse has heat sinks on the ram separate from the rest of the cooler and it has way better thermals.
I heard the AIB card's memory doesn't over clock as well. only an additional 25MHz.
I have an rx5700xt and it gets to 100c i know its not bad so im going to msi afterburner and turning these fans all the way up. Im going to update if it gets cooled or not
whoever liked my comment reminded me to do an update. Thanks. Here's your update. So I ran into 3 problems that made my GPU and pc parts overall hotter than they should have been. So first, my pc case is the NZXT h510 so it was going to be somewhat warm no matter what. I was thinking about buying a new case but I knew there was a way around it. Luckily there was. Anyway back to the update, I had only 2 fans and they were both exhausts. Which was bad. I changed one to be in the front blowing air on my GPU. my second problem was that the fans weren't running at their full potential. I turned them all the way up to blow on max. I forgot what the rpm was [sorry]. My last problem was that my GPU fans were spinning pretty slow. Around 20-40%. Overall, making these 3 changes brung my temps from 100 degrees Celcius. to around 50 degrees Celcius. I hope this helped. Have a great day :)
How's your card holding up
My card still kicking since 2020. Never repasted. Fan curve makes a lot of difference. AMD just wants us to tinker on its products.
i got a i9 9900K for 500 euro really happy
So, while playing Flight Sim I get junction temperature to 85 max, is that ok?
I agree that AMD should have kept this junction temp hidden. I personally like being able to see it, but sadly there's a horde of morons and fanboys out there using it to keep the "AMD cards are hot" meme alive. I even saw a youtube tech reviewer with over 500k subs comparing the 5700XT's junction temp against the 2060S's normal (edge) temp :(
They are proving to you that they have an agenda.
Name him and we go roast him.
I'm getting 110°C GDDR6X temps on a RTX3080. This isn't fine right?
Does it still work 🤨
@@SwayerThorSofc it would
"Up to..." Is the key phrase. 110°C is the maximum temperature. Sadly this video title is reinforcing the idea that the GPUs are hot.
haha your intro made me like and sub. Legend
ngl this help a lot thanks
Thanks haha, I was worried about my new card hot spot being 105c when I play games in 4k
Its now an issue with a lot of 3080/3090 due to high VRAM Tjunction.
Would someone be able to just put sensors all over an nvidia die to get a "junction" temp for it somehow? Just to clarify to the world what's going on
this video seems pretty relevant today lol
It really is
Should I be paranoid hitting mid 80s with junction temp of 110 ? . I get scared to even see if the thing is going to hit core of 90 so I’ll just close the application for a bit
So if my brand new Red Debil RX 6700xt have 90-95C hotspot temp in AAA games on ultra/high settings. Thats mean my gpu is broken? I ask because i dont know what to do now.
My hotspot hits 110... I have a water block I guess it's kind of an issue with 6000 serious
It makes my card throttle..
Not an apple fan : T-junction is the hottest spot from around 25 hot spots inside the GPU die and is there to monitor the silicium melting treshhold of 160 ° C and in higher quality tranzistors, like some AMD cards have, at around 190 ° C. It for the LN2 extreme overclockers to monitor their GPU-s before the melt the GPU core from within.
110 ° C IS COMPLETELY OK AND COOL ENOUGH !!!!
Actually no, junction temperature can go as high as 120ºC, the ideal is to keep them lower than 100ºC
@@MrGlitchDogePE wrong buttercup, you are repeating illiterate ideas from other selfproclaimed youtubers. You are misinformed and illiterate yourself.
Junction temps are as I have stated. 120°C is just the first preemptive treshhold so that laiks like yourself wouldn't go higher and damage the GPU at 160°C + values.
I have seen people run 5 years at 110c hotspot for 24/7 and still performs great
@@Apocalyptic47 I believe you....because the hotspot temperature is irrelevant in any GPU not until its used for LN2 OC-ing. Every modern GPU since from around 2012 had a 100+°C hotspot temperatures that just weren't reported. AMD chose to report it for the ease of use for LN2 OC-ing, that's it.
Out of curiosity, why should I believe you over AMD... I'm critical of all authority, but that being said, you need to convince me as well.... Starting with credentials
@Last Tutmos he makes a compelling point. Every claim needs proof
@Last Tutmos but nvidia cost like twice as much for 2/3 the performance, you'd have to be a certified crayon muncher to think any nvidia card other than a second hand 1080 was a good buy
Someone with credentials can lie to you or just be wrong. Not saying credentials are worthless but you shouldn't base everything on them. 2 people with similar levels of credentials can always disagree. Use your own mind and do some research if you're uncertain.
@@saunacowboy3958 nah, really? I'm a self identified anarchist (as in opposed to all hierarchy and authority and anyone or thing seeking to establish themselves as an authority has to overwhelmingly prove, beyond any doubt, that they deserve such empowerment) and a current doctoral student (molecular neuropharmacology) so it's very unlikely I'd be impressed by his "credentials" unless he had a doctorate in electrical engineering and did his dissertation on gpu architecture
@@post-leftluddite so someone being an authority as established by having credentials is meaningless due to you being an anarchist but without credentials someone's opinion isn't worth considering. I'm glad to see the education system is producing well reasoned and rational people like yourself.
I have the Anniversary model and I've went to paste (kryonaut) and washers and it generally runs about 80c on the edge temp and about mid 90's on the junction. But, during Metro the junction temp will get right on 110.... and all of this is with an undervolt down to 1102mv and +20% power limit to keep the boost clocks around 1980mhz. My GDDR6 isn't good either. Auto overclock on the RAM picks 895mhz and it's unstable. I can get it to go to 890mhz but it doesn't help much.
I would check these 2 things!, 1) Bad airflow threw your case. 2) or you still have a bad mount between your gpu die and the cooler. I would also try and see if you can undervolt some more on the core voltage, I will see what voltage I can run 1980Mhz at and will post back, might be of some help to you!. I can get about 1984Mhz at 2054Mhz/1076mV with a gpu temp of 67c and a junction temp of 82c, but the blower fan on my custom profile is up to 3000rpm which is a bit noisy, what fan speed are you running?.......
@@Chris7uk00 Bad case air flow is very probable. I'm currently housing the 5700XT in a Fractal Define R4 which is my old gaming PC. My 3700X died after a few weeks and I'm waiting on the RMA, the new PC is housed in a Fractal Define S2 which I am going to mod the front panel with some stylish looking holes with a plasma cutter I have at work. Think I might go for the same kind of look as the iCue 220T. I feel like the mounting pressure is pretty good. I really torqued down the bracket and used the washers. I got these super thin plastic ones online that are actually .05mm thick so I used two on each screw. I tried using just one on each and it was the same. The max fan speed is set to 48% which comes to 2400 RPM, I can go higher but it'll annoy me some. Core clock is set to 2044mhz (which is what stock was) at the 1102mv which comes out to 1.106 during load. I could probably take it further but the problem I run into is when I undervolt the card the game I play the most, Final Fantasy XIV, will crash randomly with DirectX errors. This doesn't happen if I leave the stock voltage alone. Even if I use Chill and run the FPS down from 60 to 70 it will still crash, sadly. I thought about taking it apart and checking to see how the thermal paste looks but I'm out of Kyronaut but I do have some Artic MX-4. It's not as good but it may be show much difference.
I can’t stand such a low ddr6 frequency on the 5700 because we pay a lot if more money for our higher frequency ddr4 RAM knowing how little impact it have on games. I’ll pay a 100usd more for a 5700 card if it has a good memory clock as nvidia cards.... Oh no!! I’d better buy nvidia and enjoy RTX and the fantastic driver support
Yes exactly my thoughts better driver support better OpenGL support. The temperature are on AMD even when we look core temperature not junction temperature I'm talking about core temperature atleast 10-15c + higher then Nvidia. I use right now a XFX RX 580 and my temperature in full load gaming are between 75-90c and this is not normal. Thats why I said f*cK you AMD I purchased now a RTX 2060 Super Aorus is coming this or next week probably next week on Monday.
@@strohhuttv8848 Awesome GPU, I'm going for RTX Aorus also. Another thing, take a look at the latest nvidia drivers because is a giant middle finger to all AMD features. And about your temperatures, I'm sure you can fix it, just buy a good thermal paste like Kryonout and take apart your GPU and also clean it.
@@arielpuig7302 I mean I can but I would kill my warranty. I think I will more likely sell my RX 580.
I'd like to see you sit down and discuss this with good ol' Steve. The misinformation really ticks me off!
the only reason most of you think blowers are dead is because nvidia made it's Geforce RTX cards dual fans.
did you not see that nvidia's higher class Quadro RTX cards are all blowers. Blowers are not low end.
It depends on the heat-sink. Some blowers have high quality (high mass) heat-sinks and some feature vapour chambers.
But I have always found that blowers tend to be noisier as there is only ever one blower fan and it needs to spin up a lot under heavy loads, thus making them noisy.
just here after my junction temperature reached 85°C while playing certain game for a few minutes, I was a bit scared lol
my chinese gpu has a month of use now, maybe I'll change the thermal paste since I have a pretty good one
OK. 110oC temperature not relevant to me. So why my fans then spining at 2600 rpm ? when gpu temp. lets say 75oC (meantime hot spot 107oC)
I got the MSI 7900 xtx and besides the power limit only being allowed to increase by 8% its been fine. Hotspot can reach 95c which is way too hot for me but at least it isn't 110c right?
I can verify that Nvidia does the same, my 1660ti won't go above 65c before it starts to hit thermal limits, granted it's only every once in a while and it's way more power limited, but it will hit thermal limits around 65c on Nvidia
As mentioned I can pin the power limit, but every now and again it will temperature throttle as well
It seems that memory is reporting the controller speed rather than the speed of the GDDR6 as done on Nvidia, with that in mind the memory is pretty fast, my 1660ti won't go much above 875 under normal conditions if calculated the same way, it's like 7k mhz so AMD would be around 7,4k if calculated the same
Great video as always
You can bet if it was Nvidia, they would have gotten the point across to youtubers, reviwers properly and if anyone big was calling it core temp, Nvidia would be on them in an instant.
Repaste.... mine hotspot temp went down by 21c after repaste with high quality paste. 6900xt
The way to deal with junction temps is to ignore them. Great that AMD has the added info, but that data should never have been put in front of the consumer.
agreed
Hey I just bought a used 6700xt and temps seem odd to me but maybe you can clear it up. Gpu temp is 63c and hotspot temp is 110c in game. Is this normal or not? I see so much conflicting information. The fans seem to run at 100% at all times while gaming to. I did remove my case glass panel and the temp did drop 10-15c so is my case just suffering bad air flow?
ughe 3600 at stock was running 90c during gaming and it isnt that hot like it is today it was hiting the advertised speeds of 4.2 ghz though but now i bet it would run at more then 90c during gaming but i applied the -100 voltage offset till the agesa update now even at this sauna of a day it runs max 75c and keeps ghz at 4.050 during that yeah im just saying ryzen 3600 ryzen 90c not very comfortable even if it doesnt stay there long no ty and seems aio is kinda worse for ryzen 3000 then a good liquid performing air cooler since the aio takes to long to disapate the heat
My Vega 56 uc is much less effective at junction temps reaching 80 C plus (often with all the other temps in the low 60's). That said, I have gotten the junction temps much higher with no ill effects. It just gets better performance with junction temps much lower. My card's sweet spot for performance is around 70-72 C hotspot. As said in this video or was it someone else there is a difference in what the silicon can handle and what is the most effective. Like the Vega 56 being able to reach 90 and not blow up ; Can RX 5700 XT reach 110 safely? Yes but your probably doing something wrong if it does and its hampering performance.
I cant see the same effects on my blower 56. Its more worth to take higher temperature to get more performance. For my card it is better to go for 220W, letting it run up to 85°C core, with an as low as possible fan speed than trying to press down the chip to 60°C, while having only ~160W. The Card will be faster and quieter with the first mentioned settings its around 1440Mhz vs 1380Mhz. That might be because the sweet spot of that blower is ~2000RPM, not to noisy and still moving a good amount of air.
@@oxxExtinctionXXO that's great but btw I have a 3 fan version and my card is only "uc" so I can better thermals to run it harder. So I get almost a constant 1672mhz with 940 on the hynix mem while running 58 on most thermals except hotspot which hits between 70 -72 under 100% stress load. Firestrike is really good but games vary from 5% on a few games to the more normal 15% or 20%. But if you get that with an overclock more power then that means all these chips act different or the cooler forces different methods
I’m just sitting here wondering how the hell its 15 degrees where you’re at while I’m sweating my ass off in 90 degree weather here in Florida
Do you know the difference between Celsius and Fahrenheit?
I had a 5500xt and now a 6800 and never seen a junction temp anywhere close to that. The hottest I could get a junction temp was 90 when stress testing the 6800 with time spy.
The junction temp being really high with a low die temp would simply be because the workload cannot utilize all of the compute units equally so one section is working really hard while the others are barely working or not at all.
and how can solve this issue to compute units equally?
@@reapergaming5916 A task can only be split up into so many bits and some of those tasks are more difficult than others. Perhaps depending on the card, it's either time to re-paste or there were defective Radeon 7900 XTX reference coolers.
@@cracklingice I'm getting gpu temp of about 70 and junction temperature around 92-95, is there something wrong with my graphic card?
@@reapergaming5916 No clue what your card is. My 3080 has a 10 to 13 degree delta between temp and hotspot temp. It may be wider if I pushed the card harder. I run it at 80% power limit to cut down noise and heat production, but I could run it much higher (and thefore much hotter).
@@cracklingice mine is rx 6700
Why not just put another heatsink on your junction?
so the memory temp of my gpu is 106c while mining is it fine or not just tell me that ?
the problem isnt the addition of the different temp info....its stupid people who dont take the time to understand what they r seeing.....but they do run too hot with the blower, which is no surprise....but could u imagine the junction temp a blower equipped 2080ti would have??
Mines doing this now. I told people it was running hot and the pro's didn't believe me lol
my hot spot is à 100°, it's normal ? i have rog strix rx 5700 xt.
Yeah fairly hot for a rog cooler but still well below high temps 110 is throttling
My gpu temps on my rx 5700xt get up to 85-ish degrees, how do I lower it and get better airflow
you made this easy to understand, they're stupid if they don't get it after this video.
Paul "get your facts correct". Buuuuut, he is Irish... so due to his lovely accent I hear "get your fucks correct" ;)