High hot spot temps on 4090 ROG Strix

Lifelite

Supreme [H]ardness
Joined
Aug 26, 2015
Messages
6,997
Long story short, I got a new case to have everything all match with my PC setup (all rog strix because apparently I have more money than sense, but I like the aesthetic).

I went with the front facing video card option, which looks great but there isn't a ton of space between the fans and the glass case (probably half an inch). I think airflow may be half the issue.

When gaming, the gpu hotspot gets up to 103C; though the other temps look a bit high but nothing I would be concerned over with 70-80C being the range.

1715013946417.png
1715013989198.png


I'm thinking the obvious solution would be to put it back to "normal" orientation; though I would really like to avoid that as I do like the look. I was looking into custom loop; though that's an expensive endevor that I wasn't planning on getting into for this setup, but not off the table. (The recent drama with EKWB also throws a big wrench) Other solution maybe if there's a good AIO cooling solution out there, but drawing blanks as most of them come with the card, not standalone.

Figured I'd post in here and see if there's any other good ideas and feedback. Performance wise it runs completely fine; it's only certain games that give me the high temps, and again, it's the hotspot that's of concern.
 
I can't quite see the front of the case but the bottom has no fans. Does this case have clear, unobstructed intake airflow? I can see plenty of exhaust. I don't think I've seen my 4090 TUF's hotspot temp higher than 82-84c ever.
 
103C is really high imo. My Suprim X Liquid has a hot spot of like 72ish? Or 70? (Mem temp is about the same, core temp is much lower) And that is an AIO GPU, so the rest of the components on the PCB just have some tiny heatsinks and a blower fan configuration. Only the core actually has AIO coverage. You have a huge heatsink on the entire PCB.

You might want to make sure that heatsink is seated properly, if you can do so without voiding warranty.

Although one other caveat is that MSI limited power to their 4090s. I'm not sure if your Strix is running in 600W mode or something.
 
Last edited:
Long story short, I got a new case to have everything all match with my PC setup (all rog strix because apparently I have more money than sense, but I like the aesthetic).

I went with the front facing video card option, which looks great but there isn't a ton of space between the fans and the glass case (probably half an inch). I think airflow may be half the issue.

When gaming, the gpu hotspot gets up to 103C; though the other temps look a bit high but nothing I would be concerned over with 70-80C being the range.

View attachment 652137View attachment 652140

I'm thinking the obvious solution would be to put it back to "normal" orientation; though I would really like to avoid that as I do like the look. I was looking into custom loop; though that's an expensive endevor that I wasn't planning on getting into for this setup, but not off the table. (The recent drama with EKWB also throws a big wrench) Other solution maybe if there's a good AIO cooling solution out there, but drawing blanks as most of them come with the card, not standalone.

Figured I'd post in here and see if there's any other good ideas and feedback. Performance wise it runs completely fine; it's only certain games that give me the high temps, and again, it's the hotspot that's of concern.
bottom intake is usually best for this sort of setup. Short travel distance for the air. Naturally moves towards the exhaust. But your case doesn't have bottom intake.

Front Intake would certainly help. I can't tell if your case has that.
However, its not ideal, because the air from the front has a long way to travel, has to resist the pull of the 360 AIO at the top---before it even makes it to the GPU. And also has to make a 90 degree turn, into the GPU.
 
Last edited:
103C is really high imo. My Suprim X Liquid has a hot spot of like 72ish? Or 70? (Mem temp is about the same, core temp is much lower) And that is an AIO GPU, so the rest of the components on the PCB just have some tiny heatsinks and a blower fan configuration. Only the core actually has AIO coverage. You have a huge heatsink on the entire PCB.

You might want to make sure that heatsink is seated properly, if you can do so without voiding warranty.

Although one other caveat is that MSI limited power to their 4090s. I'm not sure if your Strix is running in 600W mode or something.
No idea how to tell on that one. I can reapply thermal paste and such without voiding warrantees according to Asus
bottom intake is usually best for this sort of setup. Short travel distance for the air. Naturally moves towards the exhaust. But your case doesn't have bottom intake.

Front Intake would certainly help. I can't tell if your case has that.
However, its not ideal, because the air from the front has a long way to travel, has to resist the pull of the 360 AIO at the top---before it even makes it to the GPU. And also has to make a 90 degree turn, into the GPU.

Yeah, does this case have intake fans at all? Its unclear.
Yes it has intake fans, 3 120mm ones
1715039262232.png

(old pic)

Granted, the airflow is restricted by a fron glass plate. Not sure if the case fans are impacting as much, the cpu under load is around 75, GPU core temps are 70s, just that hotspot is hitting around 100

put it back to "normal" just to test, but those temps are technically within specs, just not what we would like.
I'll probably do this tomorrow just to see what changes, if anything.

Temps while gaming:

1715040310283.png
 
So right now you look like you have a lot of negative pressure in the case, along with your GPU alignment. You have 4 total exhaust (granted 3 through an AIO heatsink but still) and 3 intake. That's a bad thing. You want more intake than exhaust. Right now your intake air is probably just going right in and basically getting sucked right out, while barely even touching the GPU. The GPU fans should draw some in, but I don't think it's that much, since they're sideways and barely have any space from which to grab it. The path of least resistance is definitely just scraping your back panel and then flying right out either through the back fan or the top. The 4090 is a 425-600W part, so give it a dedicated air source.

Still, the Strix cooler AFAIK is pretty massive. Even operating at a very substandard level, it's just hard to believe that your hotspot could be that bad.

First order of business is still fixing that airflow though. This is the problem with the fancy glass cases with silly vertical mounts of the modern era. Too much form over function.

Edit: also what are your memory temps at? Use GPU-Z.
 
Last edited:
GPU core temps are 70s, just that hotspot is hitting around 100
That seems very out of character for a STRIX 4090.

IMO you have two choices, RMA it or try to fix it yourself with a re-paste. (third option is of course do nothing)

Given the actual core temp is exactly where it should be, i'm thinking a mounting pressure issue. Over 15c or so there is clearly a problem. What power are you seeing when you're at 70c on the core?
 
Actually quick question before you fix any orientation, have you tried simply running it with the glass panel taken off? That should eliminate any airflow issues and would basically be an open bench. If your hot spot is still that bad like that, then you might want to RMA as the above user says. I don't think I've ever seen a 4090 with a hot spot that bad. Doesn't matter if its in spec technically, Intel kept telling us their 14900k were "in spec". Doubly so if that's anywhere close to your memory temp...
 
So right now you look like you have a lot of negative pressure in the case, along with your GPU alignment. You have 4 total exhaust (granted 3 through an AIO heatsink but still) and 3 intake. That's a bad thing. You want more intake than exhaust. Right now your intake air is probably just going right in and basically getting sucked right out, while barely even touching the GPU. The GPU fans should draw some in, but I don't think it's that much, since they're sideways and barely have any space from which to grab it. The path of least resistance is definitely just scraping your back panel and then flying right out either through the back fan or the top. The 4090 is a 425-600W part, so give it a dedicated air source.

Still, the Strix cooler AFAIK is pretty massive. Even operating at a very substandard level, it's just hard to believe that your hotspot could be that bad.

First order of business is still fixing that airflow though. This is the problem with the fancy glass cases with silly vertical mounts of the modern era. Too much form over function.

Edit: also what are your memory temps at? Use GPU-Z.
Might be dumb, but would unplugging the back fan resolve negative pressure issues? Perhaps editing the fan curve to lower exhaust and increase intake. When I get time I'll spin up GPU-z and check things out.
That seems very out of character for a STRIX 4090.

IMO you have two choices, RMA it or try to fix it yourself with a re-paste. (third option is of course do nothing)

Given the actual core temp is exactly where it should be, i'm thinking a mounting pressure issue. Over 15c or so there is clearly a problem. What power are you seeing when you're at 70c on the core?
I'll get back with you on the power draw.
Actually quick question before you fix any orientation, have you tried simply running it with the glass panel taken off? That should eliminate any airflow issues and would basically be an open bench. If your hot spot is still that bad like that, then you might want to RMA as the above user says. I don't think I've ever seen a 4090 with a hot spot that bad. Doesn't matter if its in spec technically, Intel kept telling us their 14900k were "in spec". Doubly so if that's anywhere close to your memory temp...
I can't remember exact numbers but I did this before and it lowered temps by 10C (IIRC). I'll get more exact numbers when I game tonight.

Went ahead and ordered some Thermal Grizzly Kyronaut paste. Judging from what I've read there's a chance there's some thermal paste pumpout going on, and I'm always happy to tinker. At best, it'll solve the issue, at worst, I have new thermal paste applied.
 
I was watching hotspot temps last night while playing Horizon Forbidden West (4K max, DLSS Quality, 118fps framerate cap). My TUF 4090 was only at roughly 65% load due to the framerate cap but core temps hovered around 58-60 and hotspot around 70c (reported by HWiNFO)
 
Might be dumb, but would unplugging the back fan resolve negative pressure issues? Perhaps editing the fan curve to lower exhaust and increase intake. When I get time I'll spin up GPU-z and check things out.

I'll get back with you on the power draw.

I can't remember exact numbers but I did this before and it lowered temps by 10C (IIRC). I'll get more exact numbers when I game tonight.

Went ahead and ordered some Thermal Grizzly Kyronaut paste. Judging from what I've read there's a chance there's some thermal paste pumpout going on, and I'm always happy to tinker. At best, it'll solve the issue, at worst, I have new thermal paste applied.
Get some PTM7950 and the problem will be solved forever. Don't bother with pastes, they all have the same outcome on these GPUs.
 
Get some PTM7950 and the problem will be solved forever. Don't bother with pastes, they all have the same outcome on these GPUs.
He probably should get an apples to apples comparison by using the same game/bench before opening it up and doing a re-paste. I've done lots of cards myself but it's still a job I'd rather avoid if possible.
 
Despite what mr. Jayztwocents is touting, Kryonaut is not something that holds up for a long time. If you think you have pump out now (bs btw) you haven´t seen the low viscosity of that pink stuff yet.

And opening a gpu to repaste it just because it may have pump out is bonkers until you haven´t tested the gpu in its intended orientation.

Since repasting better than the factory did, is usually almost impossible.

And does your gpu even use paste?

Isn´t it rather PTM? That stuff sticks like a Mofo if you try to remove the cooler on ambient temps. So do it quickly, straight after a gaming session.


I would do the sane thing and modify the 90 degree bracket so that the gpu sits closer to the motherboard so that the gpu can finally breathe.

Have a closer look at the bracket. I ´ll bet you will find a solution to modify the bracket or to switch it out with a third party one.

Etsy (of all places) i where i got my last 2 gpu brackets from.

fyi
My personal results show 11-12% worse heat dissipation in that orientation.
But that´s with a "friendly" gpu cooler. msi for example always does worse due to a different heat pipe used. Its internal wick structure does not work upwards as as good as others do.
 
Last edited:
Get some PTM7950 and the problem will be solved forever. Don't bother with pastes, they all have the same outcome on these GPUs.
Never used the stuff so figured I'd stick with what I know. Plus hard to tell the real stuff from the not on amazon.
He probably should get an apples to apples comparison by using the same game/bench before opening it up and doing a re-paste. I've done lots of cards myself but it's still a job I'd rather avoid if possible.
Yeah, still doing that tonight, but if a repaste is a step eventually, might as well get the stuff now instead of having to wait around on it later.
Despite what mr. Jayztwocents is touting, Kryonaut is not something that holds up for a long time. If you think you have pump out now (bs btw) you haven´t seen the low viscosity of that pink stuff yet.

And opening a gpu to repaste it just because it may have pump out is bonkers until you haven´t tested the gpu in its intended orientation.

Repasting better than the factory did is almost impossible.

And does your gpu even use paste?

Isn´t it rather PTM, which sticks like a Mofo if you try to remove the cooler with ambient temps.

I would do the sane thing and modify the 90 degree bracket so that the gpu sits closer to the motherboard so that it can finally breathe.

Have a closer look at the bracket. I ´ll bet you will find a solution to modify the bracket or to switch it out with a third party one.

Etsy of all places i where i got my last 2 gpu brackets from.
They use the honeywell PTM on everything but the dye, which apparently use thermal paste. I've seen on their forums that some have found poor contact from the heatsink to the dye, many linking back to this reddit post: https://www.reddit.com/r/nvidia/comments/116s1o9/asus_strix_4090_oc_repasteputty_mod_temps_now_way/
Besides, my wife needed some for swapping out the AIO cpu cooler on her machine, she picked that particular one just because it came with the cleaning/application kit, though if jayztwocents is touting it, maybe should have used something different, that guy has been off the mark too many times to trust his recommendation.

It's a last resort but definitely beats the "throw your 8800gtx into the oven" fix of the past lol.

As for the bracket being closer, I would have done that but the PCI extension cable isn't very flexible, and the angle required (2 sharp 90 degree 'turns') would end up putting unnecessary strain on the connectors.
 
Oh and the according to my PSU, around 670 watts power draw
 
The case is bad, that's clear, but you're still getting a 20+c delta between core and hotspot, nearly 25c. That's a problem you're not going to solve with more cooling.
 
According to GPU-Z, you're not really pulling any obscene amounts of wattage. You're actually using less than my MSI Liquid is when I'm playing Cyberpunk (~480). If you were, your hotspot temp would probably climb even higher.

I reckon this is just a slightly defective GPU. Maybe the cooler is loose, or they just didn't put thermal pads where they should have. Maybe both. Maybe just check to see if all of the screws that hold the backplate on are nice and snug, but honestly I would consider RMAing to ASUS. Reason being that you tampering with it could be a reason to deny claims later, regardless of what they say verbally right now. I think there was a 7900XTX brand (low end ASRock or Powercolor) that had a similar issue this gen. It tended to ship out of the factory with the cooler being a bit ill-fitting.

Of course you'll be out a GPU for a while, and it is, after all, ASUS RMA of all things. I'm not sure if they've gotten better but they used to be a bit notorious. Your call. Regardless of your card's orientation, it's pretty clear that this thing shouldn't be hitting those temps normally, and you probably don't want to have to change your entire case out (which was expensive) because of a card defect. You could also just ride it out, because it's technically reasonably in-spec. Up to you.
Might be dumb, but would unplugging the back fan resolve negative pressure issues? Perhaps editing the fan curve to lower exhaust and increase intake. When I get time I'll spin up GPU-z and check things out.

It would make it a bit more even, but wouldn't really give you positive pressure. You could try simply changing the rear fan to intake instead of exhaust. Not sure if it would help, but it might. It would give a bit of counterflow to the intakes up front, which might cause more of the air to circulate to the hungry GPU down bottom. The exhaust would probably still find its way out. I don't have a fluid/aerodynamics simulator in front of me to verify lol. The interactions inside of a case get a bit complicated, intuition will only take you so far.
 
Last edited:
It's a $300 case fellas, pretty sure it has front fans.
It's not the lack of fans that is the issue, it is obstructing the fans' ability to draw in outside air. InWin, for example, is notorious for making expensive cases where the intake is covered by glass. I don't know why their designers think this is a good idea.
 
It's not the lack of fans that is the issue, it is obstructing the fans' ability to draw in outside air. InWin, for example, is notorious for making expensive cases where the intake is covered by glass. I don't know why their designers think this is a good idea.
Yeah InWin has always prioritized looks over function.
 
So right now you look like you have a lot of negative pressure in the case, along with your GPU alignment. You have 4 total exhaust (granted 3 through an AIO heatsink but still) and 3 intake. That's a bad thing. You want more intake than exhaust. Right now your intake air is probably just going right in and basically getting sucked right out, while barely even touching the GPU. The GPU fans should draw some in, but I don't think it's that much, since they're sideways and barely have any space from which to grab it. The path of least resistance is definitely just scraping your back panel and then flying right out either through the back fan or the top. The 4090 is a 425-600W part, so give it a dedicated air source.

Still, the Strix cooler AFAIK is pretty massive. Even operating at a very substandard level, it's just hard to believe that your hotspot could be that bad.

First order of business is still fixing that airflow though. This is the problem with the fancy glass cases with silly vertical mounts of the modern era. Too much form over function.

Edit: also what are your memory temps at? Use GPU-Z.
It looks damn near 1:1 to me. 3 unobstructed intake fans vs 1 unobstructed exhaust + 3 going through a radiator and the PSU fan. I’ll bet total cfm in is very close to cfm out. Also, not all cases are designed the same. Some work better with a vacuum some with positive displacement.
 
So what you all are saying is, I should say fuck it and do custom loop water cooling :ROFLMAO:

But seriously, Asus has self repair stuff that doesn't void the warranty; so replacing the paste or whatever isn't going to void that from everything I've read, unless I do something stupid.

I'm almost at the point where custom loop does sound like a decent option just because I'm already taking things apart anyway and I've been wanting to try it...just might cap framerates for a while on certain games or reorient the GPU until I feel like paying for it.

The other side is, with the drama around ekwb I don't even know where I should start with picking up that pack when I'm ready. Many attest toward Corsair's kit but I've not the best experience with that brand, but of course, that's not necessarily related to their cooling products.

EDIT: also any option around sending it off is a no go for me just beccause that's a long time going without a GPU and I'm a heavy PC user.
 
asdf.gif



Work and life got busy there for a while, but this past weekend got GPU reoriented (normal orientation, without the extender cable and side mount).

Above's the current temps. it certainly SEEMS like it's running SO much better, even my PC overall (though I did fix other issues such as a duplicate microsoft game input service, etc.) That hotspot - Core Temp delta is still rather large though.

Thinking it's safe to say thermal contact on the hotspot isn't great, so some GPU surgery might be in order. So far it's not causing any issues in gaming and such, so might wait till I get around doing a full custom loop setup, depending on if that 5090 is coming or not (and the price...).
 
I agree.
Either brutally fasten the 4 middle screws (not removing due to spaceing but destroying the springs) or ditch paste alltogether.

🤔...and don´t forget that liquid PTM has to dry for 12 hours before assembly.

;) So why bother?
 
Last edited:
Back
Top