HOT VRAM! Is YOUR Nvidia RTX 3080 / 3090 Thermal Throttling? DO THIS!
ฝัง
- เผยแพร่เมื่อ 13 ก.ค. 2024
- #nvidia #rtx3080 #rtx3090
In some use cases of intense VRAM usage, such as rendering, 3080 and 3090 GPUs such as the Founders Edition may exhibit cool GPU core temperatures but 100% Max fan speeds due to hot VRAM. Let's discuss some possible fixes such as thermal pad upgrades.
Remember to Subscribe and smash that like button if you enjoy it!
00:00 Intro, Smash that Like button!
00:35 Is your Nvidia RTX 3080 or 3090 Vram thermal throttling?
01:02 While rendering a video, my fans went to 100% BUT the GPU was under 70C Core
02:08 Fan speeds were ramping up according to the hot VRAM, HWINFO update can now see the Tjunction max temp
02:38 The issue does not happen in most games, unless it's intense Ray Tracing games, but rarely happens
03:08 Issue appears when VRAM is being pushed - such as rendering, mining, and not as much in gaming
04:03 People using rendering apps and mining were the first people to run into this issue
04:31 Igor Labs had writen about this issue months ago
05:33 Which GPUs are effected? 3080 FE and 3090 FE for Sure
05:53 Thermal Pads on the GPU are the key to this
06:04 EVGA 3090 FTW3 shows in Precision the Vram temps, they seem OK under 80c, so these not affected
06:32 Air Cooled 3090 FTW3 62c GPU Core, vs Hybrid Cooler 3090 FTW3 52C GPU Core, VRAM not as affected and maybe 3-4C difference
07:30 Is your GPU Affected? What are the solutions and what can you do if you're having this issue?
07:59 Another solution, undervolt, reduce power or memory speed depending on your app, may not reduce performance that much
08:17 Running fans on high help reduce VRAM temps, but at the cost of high noise
09:08 Upgrading and replacing thermal pads on the back of the 3080 or 3090 may be helpful
10:08 Water Cooling your GPU will also help - even active cooling backplates for the 3090 are coming from EKWB
10:31 Conslusion - if your use case ramps up the VRAM temps and fans, consider thermal pads or other options to improve it
Amazon Affiliate Links:
Intel 10900k amzn.to/2YZXsr5
Alternatives:
Intel 10850k amzn.to/32HTRQR
Intel 10700k amzn.to/2R6p2yF
AMD Ryzen 3800XT amzn.to/35FBVYR
Alternatives:
AMD Ryzen 3800x amzn.to/31dcppS
AMD Ryzen 3900x amzn.to/2GusQrh
AMD Ryzen 3950x amzn.to/32Uda8h
Asus Z490 ROG Strix amzn.to/35dlBOS
Asus Z490 Apex amzn.to/33znJhn
Asus X570 Rog Strix amzn.to/35dvcFs
Gigabyte x570 Master (Should be $359 or so) amzn.to/39OFqMD
MSI Gaming Pro Carbon x570 amzn.to/3gYRNaU
Asus X570 Formula amzn.to/3kpnPyN
MSI X570 Godlike amzn.to/2FC5fVt
bequiet! Dark Base 900 amzn.to/2E8USaV
bequiet! Dark Base 700 amzn.to/35MujDQ
bequiet! Pure Base 600 amzn.to/39WRdbX4
Fractal Design Meshify S2 amzn.to/35K6Umo
Maingear Vybe amzn.to/3hzhGy9
Lian Li Dynamic amzn.to/33ymCP6
Lian Li Dynamic XL amzn.to/3kqyFoc
Hyper X 16gb ram amzn.to/2D96Sc0
G Skill Neo RGB 16gb 3600mhz amzn.to/2FALRIs
G Skill Trident Z RGB amzn.to/3mrrRsl
Corsair 1TB Gen 4 NVME amzn.to/2FOMqOo
Gigabyte 1tb SSD Gen 4 amzn.to/3hWC9xr
Evga 1000W amzn.to/2Ryucn1
Evga 1000W P2 amzn.to/33AI2Ll
Asus ROG 1200w amzn.to/33vU8W4
Silverstone 1000w PSU amzn.to/33ywPec
bequiet! Straight 11 PSU amzn.to/33yn9AA
Dark Rock Pro 4 CPU Cooler amzn.to/3kvuJmp
Noctua CPU Cooler amzn.to/3mtilou
Recommended Monitors:
BenQ 240hz Monitor amzn.to/3iCE4YJ
Alienware 240hz Monitor amzn.to/2RxXPoC
Asus 1440p 165hz Monitor amzn.to/32znV0E
Acer 1440p 144hz Monitor amzn.to/3mscG1Y
Acer Predator XB273K 27" amzn.to/2ZJnyPq
ASUS ROG Strix XG27UQ 27” amzn.to/2EcAgyI
Acer X34 Monitor amzn.to/3iFSBTy
Alienware 34 inch Ultrawide amzn.to/2RBmxnS
Follow Classical Technology on Social Media:
Facebook: classicaltech
Instagram: / classicaltech
Twitter: / classicaltech3 - วิทยาศาสตร์และเทคโนโลยี
haven't noticed my 3080 thermal throttling because my 3080 doesn't exist
How about now?
@@antuanos I have a 3080 Ti. I wonder if this applies.
This doesn't apply to you then. lol
Yeah, it sucks.. my 3090 is hot though.
@@Hargak rip
I've noticed that the temps have definitely increased on my air build proportionally to higher MSRP and out of stock situations. Voltages, however, are some of the lowest I've seen.
You killed it on this video,,, really really made, incredible organized stream and the time stamps nice
have you tried this test in flight sim 2020 in vr? I ask because i’ve seen vram usage up to 14gb.
Thanx for the tip - I did smash this like button and my GPU temp went down 1,5C. What the cool feature you have here.
Oh yeah!
Ya I noticed this with my 3080Fe too a while ago. When playing games I would have super silent fan speeds. But as soon as I started mining, my gpu fans would boost to 100% for like 10 min, then go back down to around 60% even though my GPU temp was around 65C. At first i thought it was my mining program taking control of my fans, but upon looking at GPUZ, the perfcap was telling me I was thermal throttling. It was then I realized my VRAM was running hot. My solution was super simple. I just always have my GPU fans running at 70% speed. I don't really mind the added acoustics. When gaming, my gpu temp never rises above 65c. When mining, I now see my GPU temp at around 51c and no thermal throttling. If you don't mind the extra noise this card can run super cool.
Yeah. The fans at higher rpm definitely help if you’re ok with the noise
Hey im mining with the 3090fe my gpu is around 50 degrees but my vram is at 100 degrees. May i know your vram temp?
What sleeving cables do you have on that 3090?
Sounds simple, but if you can get a fan pointing directly at the backplate it will help considerably. That option is only available for the very creative or those with GPUs on risers
Definitely, airflow seems to help a lot - for example If I keep the front panel open on a Fractal Design Define case, it lowers the temps by a lot vs having it closed
Mine has just started doing this after bootups. Had no issues for three months until now. After I boot, everything spikes, the temp, usage, and fans go way up, just sitting at desktop. After it does it once or twice, it's usually fine until I restart. I can't for the life of me figure out what changed to cause this all of a sudden.
Where can I get the original 3000 series white fabric like thermal pads please?
What is 3090 VRAM max allowed temperatures? You mentioned 110C at the end of the video - do you have source of this info? Thanks
That is the T Junction temperature, 110c. Starts to throttle - you can look up the specs for the 3090 and will tell you. Also HWInfo64 to monitor it
I'm a bit confused with this. Aren't we referring to the memory junction temperature here and not the actual VRAM? I have a RTX3090 FTW3 and it records a 96C Tjunc. The memory chips are just fine at around 62-66C, but my understanding is that all 3080's and 3090's suffer the Tjunc problem where it's running at 110C sometimes. I get my 96C by underclocking the GPU by 300 and reducing power to 70%.
👏 Great work with this video Tiago. I wish I saw this before I bought my 3090. I might have skipped buying it, but sometimes until I test it myself, I don't really see how bad it can get. 100deg is a no go for me, but even at 90ish for the vram temps which I still have to manage by adjusting the clocks/voltage, that's just unnecessarily high, like why am I paying $1800 for a GPU and getting 90deg+ temps on the vram. And there's always the potential for games to boost and push temps.
I found while playing with MSI Afterburner, the fans are based on the GPU core temperature so even though the vram is 90+, the fan curve only reacts to the GPU core. The real issue, a design issue, is that its double sided memory on the 3090, that's what I think, so there's no escaping the fact that its more difficult to cool the 3090 vram than the 3080.
Definitely, it runs very hot indeed. after I water cooled my 3090 FTW3, the vram temperatures as indicated in HW info did go down to the 80s C
The founders edition seem to run even hotter, with 100% fans they can easily stay 106 to 108c if doing vram intensive tasks
I've seen your videos too, great work!
@@ClassicalTechnology What water cooling solution did you go with? 80sC sounds pretty good right now! I have a Lian Li Galahad 360 for the CPU, so it would have to be a 240mm AIO for the GPU to fit in my case... I can't believe it has come to this. Ill prolly just hook up a case fan first on the back to see how that goes. 😄
Btw thanks for your kind words, means a lot coming from you! 👍
@@JimRPG ekwb 3090 ftw3 block, and a bunch of radiators- I did a video with the build recently, jayztwocents had it on his channel too and liked it 😁
@@ClassicalTechnology Thanks Tiago! I'll look into it. (I need to catch up on that Jayz video too.) 😁
What app is used on 6:38?
While I've NOT noticed any throttling I HAVE noticed that the backplate of my ZOTAC 3080 gets hotter than any GPU backplate I've ever felt. Swapped my rear case fan to function as an intake which blows right over the GPU backplate and it feels much cooler now.
Did you change the fan or change the functionality of it ? and what's the temp diff you noticed?
Is this more an engineering issue? Or can it be possible that nvidia can do an update to tweak the vram? I’m about to ditch my brand new 6900xt as I get stutters in couple of my main games. So I went with a 3090 👍
What are safe VRAM temps?
TJmax is like 95C, but 100-102C is operating fine. May pull down the lifespan, but fk it, going on ebay soon heh
I have a 3090 strix oc and damn those vram temps are disgustingly hot. I usually top out at 106 celcius, but I am too afraid of taking my card apart and replacing the thermal pads. I have head placing copper heat sinks on the back plate over the vram is this a good option instead?
You can make sure to get some good airflow in your case too, it can help
i also found this similar issue with my laptop, so this laptop has ryzen 5 gen 5 and i got an integrated gpu of vega 8. When the vram reach 95% load on the vram the laptop would be very hot and the fan would sound so loud, that often happens when im gaming, though the games i played are fairly light such as roblox, and often happens even when im doing an online meeting
My 3090 was hitting 110c and throttling. I used Afterburner and put it to 60% Power Limit, it went down to 86c. Is 86c OK and safe?
Thanks man, just found this out. Exact same problem.
I wasa working on an other issue and put my hand on the gpu. 3090. man that was hot.
Then I started improving cooling.
It stays on about 60 in idle??? About 70+ in game. Looking with a thermal camera.
My FTW3 3090 stays under 92C and usually is around 88-90C. I just keep the mem really low, power under 65, then you’ll get decent junc temps. Also it’s in open air case. Makes a world of difference
What are your settings?
I also noticed this issue with my FE 3090. I just stuck the corsair water block on it and added it to my loop. Now the mem temps stay in the 60s during gameplay and rendering videos. I want the EK block but I couldnt wait any longer. :) Nice video man.
Thank you! I had a Corsair block on a 2080, cool stuff. The ek one is awesome too. How was the Install process?
@@ClassicalTechnology install was easy. The corsair block came with all the thermal pads pre installed as well as the thermal paste on the die. Was actually easier than most of the ek blocks I've used.
Did you put an aio on it??
Is this possible with the corsair hydro 100i or 115 i?
msi 3080 oc 3x fan is it ok to mine 24/7 at 100-104 temps vram normal gpu is 47
Is that hybrid cooler embargoed or something? Throw a little more packing material on it. I think I may have seen a part of it.
Lol I’ll do more soon 😁
Keep bringing us content. You the man
no direction on how to change the thermal pads, or links on how to do it? I dont understand the point of this video
it voids warranty on some companies.
the point is to make money, look at all the links he posted.
@@mini-337 Manufacturers may say this, but it is illegal. Apple, Samsung, etc get fined millions of dollars every year for saying this. Unfortunately, they make more money from repairs/new devices than they lose in these fines.
www.ftc.gov/news-events/press-releases/2018/04/ftc-staff-warns-companies-it-illegal-condition-warranty-coverage
Phenomenal video. Thank you so much. LTT forums got me here. One curious question. You mention EVGA FTW 3090 version. I have XC3. Do you think both version will have the better vram temps like you found in the FTW? I didn't unbox mine yet and wonder whether I should return it or not. This is a big deal to me.
Glad you like it! Not sure on the XC3, but I think for the most part you should be ok, may have to run fans a bit higher, but would have to test and monitor the vram temps to be sure
@@ClassicalTechnology Thanks for the reply. I have a bit of buyers remorse TBH. It's why I haven't opened it yet. I need to read more findings on the XC3 and hopefully those that own the card will start checking and posting results. Tom's Hardware did give fair warning to miners that at these types of temps, 24/7, the cards are likely going to fail. I would imagine you concur with Tom's, but I'm wondering for miners what is a safe threshold for 24/7 scenarios. I'm not planning on mining mine, or at least not to any extent knowing what I know now. A lot of miners could be screwed on reselling their 3080 or 3090 cards. It should become quite clear on how mining will shorten the life span of the 3080 and 3090 cards. With the 3080 being the "hero" up until now, I'm wondering how this news will shake up the miners. Flood the market with 3080's and get money back while they can? Could be quite comical!
@@goldeneyeforevercom So what did you end up doing? I just bought a XC3 Ultra for mining and feeling remorse after reading this D:
@@cal-king I did return it. Too big of an investment for me. Not opened so no hassles at Best Buy.
I got a watercooled 3090 running at 98c VRAM, was thinking maybe needed an extra radiator. Though 98c doesn't seem that bad -I think?
The hybrids really kick in when you are running two of these in a system. They are too close to each other when using the stock fans but bypassing them with AIO you get enough clearance AND the water cooling ...!!!! By the way (2) FTW are too wide and blocks airflow and the (2) XC3 fits well but gets too damn hot (hard to get the AIO for them). Undervolting sounds like a great solution!
how do u have so many founders editions omg
getting 102, 104 even 106 degrees in my RTX 3080 ti FE VRAM, in an open case (Thermaltake Core P3 wallmounted) with max possible airflow, no other source of heat pointing to it. But is in warranty so I don't really want to repaste and repad... any idea? 😅😅
Do I need to worry about burning out my 3080 RTX Founders Edition if I am mining with VRAM temperatures between 102 and 108c Maxiumum temp? GPU temp is staying below 45c. Fan speed is 100% and Loud AF
I’d try to replace thermal pads, there FE are not good for mining out of the box
You can get a pci express fans that blow right on the video card.
Won’t help as the backplate hinders the backside of the card to get cooled
I haven't noticed this on my 3080 Founders, but I haven't done a lot of Ray Tracing with it. Good information to know. My air cooler is very close to my 3080 in my mATX board so putting anything on the backplate is not possible.
Yep keep an eye on the temps
I repad and zip tied a socket 7 heatsink on the backside. Temps never go over 80c. Gotta do whatchu gotta do.
I have the same problem with my 3090's FE . GPU temps are 59 & 69 DegC but checking the Perfcap is capping due to thermal. How do i check the vram temp?
Only some programs will show it, I use GPU Z
Same issue with my Gigabyte rtx 3080 OC. There has been an update to HWinfo64 version 6.42, now it is possible to measure the VRAM memory Temperature, it turns out the vram was hitting over 110+ degrees celsius while mining... Adding thermal pads might be a solution... The settings is under the GPU section called - MEMORY JUNCTION TEMPERATURE.
It’s juncture temp which isn’t the same as normal temps. Its just that the new silicon has more sensors now and can tell us more than before. It’s always been the case with the previous model cards we just didn’t have the info before. You’re card won’t melt bro it’s all good.
Also all of you voiding your manufacture warranty doing that
my 3080 FE was seeing arounf 88 to 90c max when doing heaven benchmark. After installing 1.5mm Thermalright thermalpads (12.8W/Mk), my temps are now always under 80c all the time.
i have 3x 3080´s they are like 43ºc but the Vram its stuck on 110ºC
So if i change the thermal pads, the temps will drop?
I've seen people online reporting success with changing the thermal pads.
does change thermals process cause the product warranty to be lost?
It can depends what indicators the company uses to tell if it has been open like tamper screws etc
I have a Aorus RTX 3080 Master I’m just going to game with it probably around 1440p max if anything. But seen in the mining community that they overheating and that a teardown revealed cheap thermal pads.
What a shame to have to tweak our self theses pricey cards ...
Yep, my "3080 Memory Junction temp" is just getting higher every day for the same workload, I am thinking of just adding some CPU thermal paste to the existing video card pads, is that ok to do? Bcos adding whole new pads seems like a lot of mess and homework.
My god THANK YOU SO MUCH FOR THIS MAN i had no clue what was going on now i know
Just noticed, your system with the EVGA 3090 has the radiator upside down. The pump should not be the highest part of the loop.
I have a 3090FE at bone stock settings save for a custom fan curve, and simply playing most "demanding" games at a 5120x1440 resolution makes my VRAM's junction temperatures peak at 102.0°C with fans at 50% (peak core temps fluctuate from 58°-61°C). Granted, I have been lucky so far, with no major thermal events since buying the GPU at launch.
Just wish there was more information on what's considered a safe operating temperature (Since HWinfo states that 110°C is when it starts to throttle, and some sources say that 120°C is when damage will certainly occur). Might just try the heatsink method since I am not too confident on opening a GPU.
Doesn't help much that the safe operating temperature for GDDR6X memory is "0°C≤T꜀≤+95°C." T꜀ is the case temperature, of course, not the junction temperature.
+95°C? Is GDDR6X "safe" at over 95°C?...
Thanks for the comment- that’s what we’re trying to figure out too, what’s a safe temp. I would guess anything below 110 is ok, and the lower you can go the better
I think we all need to relax and realize the juncture temp isn’t all that serious as we’re making it out to be.
@@capcomike can you explain ehy?
Mine runs around 102-104 it’s frustrating when the rest of the case has really good cooling. I’m still not sure which thermal pads to use and replace its really annoying that Nvidia didnt even install them correctly in the first place. They knew in advanced how hot these cards were going to run and didn’t properly address memory cooling
haha, with a Bykski block and a little fan over my backplate I get no higher than 44c on the core and 78c on the memory junction. (w/85 pwr limit, +1500 on the memory and -512 on the core), efficiency
I remember before I water cooled my 3090FE. During all gaming sessions, GPU fan speed reaches around 2100rpm and sounds like a jet.
Yeah it can for sure!
@@ClassicalTechnology BTW, did you pre-ordered EK 3090FE waterblock? They claimed the backplate is "Hybrid cooled"
Lol, try 3 fans at 3300 RPM, it’s pretty obnoxious
@@chincemagnetAt least I was a Radeon Vii owner😂
@@hansyang1697 my block arrives Friday, finally!
I use the FTW3 3090 for gamming only. I cap the max power to 65% from Precision X1. Now my tjuntion memory temps is a solid 70
I bought a pretty good prebuilt with a 3080, but the damn thing got so damn hot the computer kept cutting off. Had to downclock the CPU and GPU just to play games. I can take the performance hit, what’s the point in max performance when your computer keeps cutting off due to too much heat?
Sounds more like a power issue, the PSU is probably not up to scratch. Should maybe take a look at your case air flow also.
@@Moonmonkian it was a Skytech with a 3080 on Amazon was $2k with a 850w PSU, so idk
Does anyone know if these thermal issues also apply to the 3070 fe? Wondering if this might only be related to gddr6x ram.
Seems mostly for gddr6x
Only for gddr6x. 3060ti and 3070 are not effected
My 3080 Strix runs cool and quiet while clobbering 1440p. It’s an absolute beast.
Nice
You got lucky and it seems like they installed thermal pads on the VRAM like they should have. Unfortunately sometimes aib's don't always do this or installing very cheap thermal pads. Definitely been an eye opening experience as I've never experienced this issue with a prior 1080 I own.
Uh, unless there are thermal pads installed between the back of the pcb and the back plate (or hot plate) then installing a heat sink on the back plate will do nothing for the temps lol. Without a thermal pad, the back plate doesn't contact the pcb (for obvious reasons).
Where did you find a 3090 hybrid cooler? I really would like to swap my stock fan 3090 heatsink with one of those!!
***update*** I thought it relative that I add my current temps:
Running COD or other AAA titles (3090 ftw3 ultra) runs around 89-96c, when mining it immediately hits 110c and begins to thermal throttle. (Core is nice and cool at like 54-57c, its the VRAM that are really throwing a fit after previously being stable around 88 until this last week when I finally took it out of commission until I can get a fix.)
I got that one from EVGA, they sell it
Do you have a promo link to get contributor credit if I would like to buy one?
On water my 3090FE with a Corsair water block while exporting a 4k video on Premiere gets to 35C on the core. I used a Fluke thermal heat gun on the back plate and the hottest part of the backplate was 47c close to the PCIE socket.
When I play games at 4k the core temp gets to 49-50c with an OC.
What size rad are you useing?
@@dallasbroadway8808 two 360mm 40mm thick.
Fairly certain I am having this issue, unfortunately. For such an expensive card, it's really unacceptable in my mind, software should not cripple the hardware under normal circumstances. When we talk about removing the backplate and adding thermal pads which side of the card are we talking about? The side that says 3090, o bare side. To be clear, I understand we apply the pads inside the card, but the card has multiple backplates.
92 degree vram max in Cyberpunk 2077 max out. Is that safe or do I need to worry about degradation?
Should be fine at 92c, 100 and above is a bigger problem
6:31 it prob doesn't help with temps either when your fans are sucking in hot air from the radiator into the case lol... oh and lets not get started on the orientation of the radiator tubing on the bottom (gasp!)
I have ordered some thermal pads to stick on the back plate however in the UK I have been warned this may void the warranty. Pretty poor if so as its hardly a risky mod.... but despite paying £20 on thermal pads i am reticent to void the warranty on my £1400 GPU. I tried to get intouch with NV customer support live chat but gave up after waiting 30 mins. Currently have the side of my case open, with heatsinks on the backplate and a fan and still hitting 106 degress now my room is warm. its only gonna get worse going into Summer.
Also in UK, have the 3090 FE so reluctant to open it up, so putting a heatsink on top of the backplate has helped? What heatsink did you use?
@@nickf619 I bought a 4 pack of small heatsinks from Amazon for £7 and then combined with my old sandy bridge I5 heatsink and a fan from an arctic 64 cooler (a bit quieter than the intel fan). I did try the entire arctic 64 cooler and it was better but huge and heavy on the backplate
I have a 3090, I notice this problem as well. I can lose 30-40 fps on Forza Horizon 4 over not that long of a time period. It starts at 185-190 on all ultra 1440p. then after like 25-30 minutes it'll get to 135-140. gpu temps are still showing they are fine though
Thanks for the comment!
Aren't VRAM modules on the back of all 3090 models? If that is the case, there is probably no heat measurement possible on the VRAM on the back of the card? Not sure, but that's my guess. Meaning the VRAM on the backside gets very hot unnoticed, and you lose the FPS as a result
@@xXSeak ftw3 is the only card i know that you can actively keep a good measurement on for the front and back of the gpu temps in precision x1
Hey, I’d suggest checking the evga temps again in HWInfo, the vram temps in the precision software is not the junction temp, not the same thing and I’m pretty sure the evga also overheats the vram. No 3090 is safe from that issue =\
I’ll double check it for sure
Can confirm, my 3090 ftw3 ultra hybrid during mining with +1300 on memory and PL at 72% gets sub 60c mem Temps on the precision software but in hw info I get 90c. Core temp is 34c. Not too concerned as I could probably easily rma it not mentioning the mining aspect
@@luisdiaz71956 90C should still be fine though. Seems to be within specs.
@@luisdiaz71956 90C is pretty damn good. Lucky you! But for everyone else on air 100C is the reality, even while playing games.
It’s not an issue it’s just protecting your card from failing before it reaching manufacturers warranty, even under heavy use. I think FE cards have always been a lil sus when it came to cooling.
One of my 3090s thermal throttles at 110. The aluminum back plate is an inadequate cooling source.
well. i have a msi rtx 2080 super gaming x trio. the vram gets up between 95 and 100 sometimes even 104c. even on 100 % fan speed, its still gets that hot. heatpads are trash apparently
my Gigabyte 3080 has this issue, the memory can go to 114 degrees celsius while the rest of the card reads 70/80, the weirdest thing is the "hot spot" is actually lower temp than the VRAM by a significant margin.
So, my 2080 super has zero VRAM cooling. Like none at all. The card runs great though. Stays at 49c during hours of gaming. Do you think this will be a problem?
You should be fine with the 2080
Yes because it just happens that the RTX 3080 and 3090 both have *GDDR6X* ram, which is very likely the culprit for the high amounts of power draw and hence high temp issues discussed in the video. In fact, an article from Nov 2020 basically confirms this *GDDR6X* issue where Rick Bergman from AMD discusses the kinds of issues that GPU designers deal with when attempting to squeeze more performance out of a GPU - www.thestreet.com/investing/amds-rick-bergman-talks-about-current-and-next-gen-cpus-and-gpus
Relevant part of article, quoting Bergman directly (when he discusses "Infinity Cache" he's referring to AMD's rx 6000s' Level-3 caching technology) - "On Infinity Cache, it's somewhat linked to that as well, to a certain degree. If you've been in graphics for a long time, you realize there's a pretty good correlation between memory bandwidth and performance. And so typically, the way you do it is you jack up your memory speed and widen your [memory] bus to open up performance. *Unfortunately, both of those things drive up power [consumption].* "
The 2080 super is beast. You can get 44-57 hash with the right configuration.
I will also just add another comment. This situation is exactly why I'm perfectly fine with older, bigger, heavier sound muting cases. Yes, like Fractal cases. Cosmetics vs. noise? I take quiet any day of the week. If ramping up fans is how new cards need to be handled then smart case buying is an actual thing. Of course cases now are light metal junk that couldn't mute much of anything.
Take a look at Gamers Nexus video on sound dampening cases. You’re actually making the case noisier by blocking off airflow than you would have if you had a high airflow case with slower fans.
Gaming I have found the max VRAM temp I can get to on my 3080 FE is 88 degrees C, not really that bad imo.
Have you checked the memory junction temps? What were those.
What is a good Vram temp?
what are the fans on the left case ? the ones that have the blue color on
Looks like EK Vardar RGBs
@@hansyang1697 just looked it up for the first time, looks similar
@@YoYoTomer He is using EK AIO cooler, so I can confirm that's Vardar RGB
@@hansyang1697 amazing ty for helping
Yeah ek vardar RGb fans
what ram kit is that in the 3090 system?
Corsair dominator rgb
@@ClassicalTechnology i only saw two of the dims lit up and 2 other slots with different looking dims mixed kit? Or fake dims?
So this is why my cyberpunk playthrough kept tanking after 30 mins at 1440p ultra ray tracing......
This is exactly why it is best to go for the GPU with the best cooling solution, no matter what. With my 3080 Ti, I barely hitting 65 degrees GPU temp on heavy load (usualy just above 60), and have never topped 80 degrees on GPU memory junction. Stays cool and quiet no matter what I throw at it. No need for undervolting and mounting fans on top of it.
What model did you get?
Which model 3080 ti do you get?
While mining ETH I always have EVGA 3080 GPU at 50ºC and Memory/VRAM at 90ºC.
Should I worry about this? Do you think I can mine ETH for many months/years? Thanks in advance!
When i mine with you 3080 STRIX i get 106-110c VRAM temps so i stopped mining for now until i decide to either change the thermal pads or waterblock.
You should be okay. 90 degrees is good compared to many other cards. EVGA makes them good at least.
@@francistan9306 Check out these pads, excellent price and drop the temps a lot: Frosty Ice Cube Plus Thermal Pad - 12.8 W/mK. It's on Amazon though so not sure if you have access to it.
Forza Horizon 5, Extreme settings 1440p - game for 1-2 hours, Thermal Throttling kicks in. THE FPS DROPS ARE NOTICEABLE SO MUCH YOU HAVE TO STOP PLAYING. And I do have under volt and fan curve on aggressive. I think you actually have to let the fans go lower so the GPU temps are high and the GPU clocks go down hopefully letting the VRAM rest a bit.
I think its more on the fast cars where you go 200mph via the country and the VRAM is really challanged to load all the textures and spit them on your screen with these massive speeds. Slow car racing is probably not an issue :)
@Jesus is Lord THanks for the Tip, I recently bought a 3080 Ti FTW3, actually it has the same issue in Forza Horzion 5, even tho the temps on the GPU and VRAM are far better, It happens even when I force lower FPS on the game. its not the temps its something else about the game that is causing it to degrade itself on peformance over time,.... it could be the more land you cover in the game the more "clogged" the VRAM becomes and the the GPU starts choking on it. Now I have 12 GB VRAM and 32 GB RAM so... don't plan any upgrades there soon. I think the game just needs somekind of patch. I was suspecting it could be the 12900k and scheduling by win11 on the threads, but I think I can safely rule thaht out. In fact the game itself reports if you let it in that state "Streaming issue" and mentions the CPU... (but the 12900k CPU is nowhere throttling temps or load peaks) - could this still be caused by the 12900k instead of the GPU tho? Ideas welcome. thanks
I added thermal pads and better thermal paste and my MSI RTX 3080 VRAM dropped by around 30 degrees.
On my Galax RTX 3080 I did the same and probably dropped it about 10-15 degrees.
Do you have the 3080 Suprim X?
@@rpospeedwagon yes. They actually put thermal pads on the backplate (which Galax didn’t). But, they must have used the lowest quality crap they could source.
@@christopherquigley5468 Galax is the same as KFA2 right? I just ordered a 3090 OC 1-click from KFA2 and am not sure if this will be effected by thermal throttle as well since it has the "4th fan to click on"
@@riplock88 I looked it up. KFA2 looks to be the same as the Galax I have. But I live in Korea so they probably operate under a different trade name.
Installing thermal pads and using better thermal grease isn’t difficult. But it might void your warranty.
In my case, a month later, things are running well. Both my GPUs are mining at around 90’C. They were throttling at 110’C before. And likely getting damaged too.
My hunch is that the extra fan won’t be enough.
Hey man, you prob don’t remember me but I was here a couple months before the AMD and Nvidia launches. I ended up building my rig with a 5900X and a 3080 FE. The experience has been great, but a week or so ago I started mining crypto. I ran my PC for a 24 hour period to plot out the energy costs and get a feel for the noise and performance hit. Lucky I saw it when I did but I have the same issue with the VRAM temps and started to using EVGA Precision and tested like 20 different configurations to bring down the temps. As I didn’t build the pc to mine, I’ve kind of abandoned the prospect of passive income now as I don’t wish to risk failure of a card I worked so hard to obtain or void the warranty by installing pads. Are there any settings you can recommend or a direction you can point me in so I can try and strike a balance between mining rate, temps, and noise? So far I’ve only messed with clock speeds and power limits.
I don't understand how some people have thermal issues in their build and their GPU. I own a 3090 FE and have never experienced such temps and most of the time when I'm gaming my gpu usage is usually at around 95-100% and hottest it's ever gotten is 65C. Maybe these GPUs just need a good case with good airflow
I have a zotac twin edge rtx 3070 and an i7 10700k i have the card 1 day and while i play gta my card is only running at 50-80% at 1080p 165hz can anyone tell me if this is normal other games like warzone are staying at 98-99% usage
Sad thing is that changing thermal pad doesn’t help. Memory chips on the back was a pretty damn bad idea. If you only play games this issue might not affect you that much, the worst that can happen is a higher fan speed. But if you render stuff, mine or do anything memory intensive, please use the default fan curve and, if possible, point a fan to the backplate if you don’t want your GPU screaming and thermal throttling.
Why would mining cause thermal throttling. Mining involves undervolting and running the cards at very low temperatures. Full voltage is not energy efficient and therefore not profitable
I would buy a mining card any day before I'd buy a GPU that was used for rendering. They run cool and steady and are generally well looked after
@@aightm8 You are an idiot that has no idea what mining is. Murilo Bastos is correct, I have a 3080 and can confirm that this is happening even at 60% power limit and -200 core. The memory simply isn't made for mining.
@@Alexrogue Or maybe you just got a garbage card that doesn't cool the memory modules very well
Graphics cards not made for mining??! Shocking insights here from youtube user Bang Bang.
GDDR6X runs hot who could have forseen such a thing.
And nope, I don't know much about mining. It doesn't make me an idiot. It just means I'm not a peasant going through so much effort for €150 a month. I mean respect if someone is from a country where that's worthwhile
@@aightm8 lol I won the silicon lottery with my unit, I can actually run the card at 96-97 mh/s without additional thermal pads and not thermal throttling.The problem is that even my lucky card is still hitting over 100 degrees on memory, which will make the cards last much shorter than intended - even for mining purposes. Also I have no idea what you are talking about 150e per month, and I have no time to decipher whatever nonsense you are going on about.
@@Alexrogue with electricity costs in most European countries, 150 euro a month per 3080 is about right. If you're really looking at the ATH of etherium to calculate your monthly earnings that's wojak brain thinking.
It's only profitable in poor countries with low electricity costs for the most part. And I'm happy for you
What size pads for 3080 FE?
1.5mm for the front 2mm for the back i think... for the 3090 its 1.5mm all round.
For me dropping the power limit to under %70 my eth mining rate didn't drop much but my temps and power usage did ;)
My strix 3080 vram temps hover around 90-95 when mining, would this be bearable to do long-term or would I need to do something to lower them? I know Micron rates the GDDR6X operating temps at a max of 95 C, so being at the upper threshold somewhat concerns me, even if I'm within the limits.
90-95 is perfectly fine.
You should be more concerned about using shit cases with meh air flow, then gpu memory running within spec. The 011 mini is good but the first few min show cases with pretty meh intake/airflow
yeah this is an issue. i just got my strix 3090 and it goes straight to 102C while mining
How are you going to solve your issue? Thermal pad replacement or water-cooling?
That's not bad actually, lol. Blow an extra fan right on top of it, you'll be fine. Aim it right at the POSCAPS/MLCs
@@user-lu6cy7hm2t 102* is pretty easy, even at 80% ..I finally got 94-96 tjmax after months of tinkering 83% tdp, 85% fans, and. voltage lock .72, no OC just for 105 mh/s
dude i thought my auto fancurve was fked, since when i ran furmark or a gpu stress test, the fans went to 90% despite the core temps hovering around 70/69 degrees, then i checked gpu -z and hw monitor and saw taht with my custom fancurve the gddr6 went up to 110 degrees and actually throttled 10 fps down, i now use autofancurve a slight undervolt, and now im faster (12k in port royal) @1950 mhz with only going to 105 on the gddr...also keep in mind that the stress test never applie to a normal gaming session where your gpu will never be permanantly at 100% load due to cutscenes loading, and just generall lower load on the gpu
My Aorus Master 3090 hits 110c on ETH mining, had to lower power to 65-69% to keep them under 90c
How did that affecting your hashrate?
@@gregbarnes4083 I was able to get good relatively high hashrate. @ 90-100% power, Memory +1000 I would get about 118-120mhs ~ 350-390 watts(Terrible pwr usage). Vram 108-110c.
Best setting I found for my card are:
Pwr: 70%
Core: -150
Memory: -502
Fans 70% or auto depending on ambient temp.
Mh/s: 93 - 98
Watts: 240 - 260
Vram temp: 92-96c (with a spare fan direct on the backplate pulling air out 70%).
my aorus xtreme 3080 vram is running at like 100 degrees celsius... damn fug
3090 FTW3 with icx sensors, mine never really gets out of the 50s C. I can’t imagine it getting that hot unless your cards cooler is poorly designed or the fans are set too low.
Not the same thing, my 3090 hybrid ftw3 gets sub 60c as well in the precision software with icx but 90c in hw info. Core temp is at 34c. This is all while mining so its running at full load
Check it under hwinfo the reported temps in the precision are external and don’t represent the relight memory temps. I can assure you ti will be running close to 100C. Only watercoled 3090s are safe
@@luisdiaz71956 I should probably specify I pretty much run my fans maxed out all the time
Doesn’t matter, water block arrives tomorrow
ICX sensors only monitor the memory on the front of the card. Not the memory on the back of the card. Your other 12 memory chips are definitely cooking near 100C. Download HWINFO or GPU-Z and take a look at junction temps.
I just got this pc like 2 weeks ago with a 3090 x trio gaming its tunning 80 degrees celciuls
my friend and I hash with 3090s and we really struggle with thermal throttling on the memory.
Afterburner settings are your friend. 80% power
@@user-lu6cy7hm2t I think it would matter more for gaming. In mining it seems to pull power away from "waste" areas like the GPU/ voltage, the vram is only 150 watts or so
Hey man Thanks for the video and including the timestamps. However, I recommend having less time stamps and more brief descriptions. I shouldn't have to read a paragraph for the table of contents, right?
I slapped an aio cooler (corsair h80i) on my gpu backplate (Asus rtx 3080 tuf) and temps went down from 100c to 92 while mining. 94 max at max fan speed
I'm planning to add a cpu cooler on the left side of the backplate now to cool it even more. 30c ambient temp.
Just adding some info for people reading comments.
I would love to know safe temps for 24/7 operation. Any info on this?
anything below would be safe
below 80
@@subhamkumar4146 I’m pretty sure it’s been confirmed to be 95C to be safe for 24/7 operation and 110C for throttling.
@@GucciGamerTV ohh then its fine
i just turn my ac split system on high and thermal throttling goes away. seriously! if your room is hot and you start turning on electronic like a 1000watt pc, you are gonna have some serious overheating problems in your pc.
Rtx 3080 Asus tuf oc have this problem?
Yes
The reason WHY these cards (specifically the 3090) run so cool (core temps) is precisely BECAUSE the card is throttling due to memory junction temps.
No
Just got my 3090fe it’s having problems with vram buying new thermal pads getting 90mh 84 constantly kicks up to 120 on start up so hopefully it’ll solve the problem
i underclocked and getting 118mh
@@user-lu6cy7hm2t It helped a bit, but changed my thermal pads and that dropped the temps on both 3090 fe from 104 to 78/84DegC
@@user-lu6cy7hm2t I got the underclock figures from nice hash mining website i use. MSI afterburner to underclock, best thing to undervolt
What is that rgb memory in your white pc
corsair dominator rgb platinum
Corsair dominator RGB
Thanks!! It looked completely different because the dominator have text in the middle
@@ClassicalTechnology Btw i already have 16gb 3600 cl18 vengeance rgb pro.... could I use the dominator with the the vengeance?
If only i had One i could have this problem 😛
My Rtx 3090 Vram temp is between 96-104 C° is it safe? (120-124mh)Ethash
No lol. Doesn't matter what nvidia says either. 100C for a PC component ain't right. If you can fry an egg on your GPU it's far too hot.
@Bryce Czirr haha