5900x, running hot

Joined
Nov 25, 2020
Messages
30
I've got a 5900x paired with a 3080. It's being cooled by a Corsair H100i XT iCue RGB Pro. It idles at ~35-40C, with the coolant staying at ~30C. That all seems Fine. When I play Destiny 2 or Star Wars Fallen Order, it gets up into the mid-70s. All that seems normal as well, but then I loaded up CoD: Cold War, and it will run in the 80s with the coolant at 40C. The CPU temp seems concerning to me. I'm not overclocking anything. I have 6 120MM fans (3 in front, one in the back, 2 on the radiator pushing air out the top of case. I'm pretty sure I'm getting good contact with the chip. I have the screws tightened down as far as they'll go by hand. I know these chips run hot, and I'm playing Black Ops at 4K w/ DLSS and everything maxed out, but I thought I would ask here to see if this is expected behavior. The system is otherwise stable. No BSOD, application crashes, freezes, etc. That said, because of the cost of the parts and the lack of general availability, I don't want to fry my system.

I also saw this thread which seemed to indicate that this is by design.

https://www.reddit.com/r/Amd/comments/jsy8bw/robert_hallocks_response_to_all_zen_3_thermal/

Thanks.
 
if you want to lower the temps, use curve optimizer. you could set a negative 10 value on all cores and see an improvement on temps. you might be able to start with a negative 15 value even.

as far as temps go, those are pretty normal as it sits.
 
  • Like
Reactions: DrDoU
like this
have you loaded icue and turned up the pump/fans at all? maybe make a more aggressive fan curve.
 
I have a pretty aggressive curve already, so good thought there. The fans I use top out at 1400 rpm, but honestly that should be plenty considering the temperature differential is held steady at 40C-45C after over an hour of playtime. Also the coolant seems to hit 40C whether the CPU is in the 70s or 80s, so I think its hauling the mail appropriately. I have the pump running at "Balanced." I could move it to "Extreme." I tend to keep my processor/motherboard for 7 years, and just make do with GPU upgrades. I didn't want to wear out my pump, which is why I have it set to the middle level. If that's not a concern I can move it up under heavy load to see if that helps.

Also, I've got good cable management, for the record. No air flow restrictions in the main compartment.
 
Sounds like your AIO isn’t able to keep up with the thermal load. Your CPU needs more radiator/fans.
 
seems fine to me. temp on 80c range is nothing for these CPUs. They are designed to handle more. I would probably leave everything on auto and not push PBO too hard since it makes little to no difference. Use Curve optimizer if you like to reduce voltage a little. But not really necessary. I read somehwere Call of duty BO4 used Avx instructions so Cold war might be using the same thats why you see higher temps with it. Would make sense.

I have the AIO connected to Motherboard pump header. It does pump at full speed I think and that does usually help moving the water out faster.
 
Spot on. Didn't notice the cooler its 240mm. 260mm with another fan will definitely help.
I'd go 280 or 360, depending on case clearance.
Arctic is all the rage right now.

Maybe look at ML fans on the rad if you can tolerate the vibration at the gaming end of the fan curve or Silent Wings 3 if noticing your pc with a headset on is an issue.

If the OP is adventurous piecing together an open loop out of used parts, new D5 pump, may be a enjoyable learning experience.
 
That's a pretty darn high water temp and a serious concern. It won't be doing much cooling with water temp that high.

If the CPU is at 70 or 80, and the cooling liquid is at 40, instead of say, 30, which is room temp, that's not going to make a night-and-day difference in the heat transferred.

q = hc A dT

q = heat transferred per unit time (W, Btu/hr)
A = heat transfer area of the surface (m2, ft2)
hc = convective heat transfer coefficient of the process (W/(m2oC, Btu/(ft2 h oF))
dT = temperature difference between the surface and the bulk fluid (oC, F


Let's say for example that the q in the above equation is 100, then keeping all things equal, if the water temp is 30 instead of 40, which creates a dT of 50 instead of 40, that would only change the q from 100 to 125. A 25% increase in cooling capacity is certainly significant, but its not going to get me down to 70C by itself, and since hc won't change, I'd have to get there by going from a 240 to a 260, and even then I'd have to spin up the fans quite a bit to get the water temp down to 30C to create that final product.

If in fact the processor can run in the 80s without cooking itself or other parts, and the rest of the components are doing fine, then it might not be worth the cost or noise to make the change.

Again, that's assuming NDK and others are correct in that the 5900x is running hot, but its not at the point of causing harm.
 
Last edited:
Have you experimented with reversing case flow yet? I wonder if your gpu isnt warming up your mb/cpu when your running those two games and thats a factor in the increase your seeing?
 
Have you experimented with reversing case flow yet? I wonder if your gpu isnt warming up your mb/cpu when your running those two games and thats a factor in the increase your seeing?
Hard to disagree with this take as the EVGA 3080 FTW3 creates a lot of heat.
 
Normally i dont worry much about case temps effecting rad temps but if your exhausting out of your aio while your gaming on the 3080 it may be a factor as to why your seeing an increase in cpu temps if those two games happen to be more gpu intensive. You may be able to keep your cpu cooler by pulling in through the top and back and exhausting through the front or ideally moving your rad to the front which ever is easiest to test.
 
You could "reverse" the airflow by either having your top radiator blowing hot air in (I wouldn't recommend this as its bad for overall case airflow) or mount it in the front as intake (will increase GPU temps some but should see your water temps and CPU temps come down) EDIT: but again those temps are fine so don't worry about it and enjoy your computer :p
 
There are lots of good points made here by others however if we are going to help you a bit more info is needed.
I have a pretty aggressive curve already, so good thought there. The fans I use top out at 1400 rpm, but honestly that should be plenty considering the temperature differential is held steady at 40C-45C after over an hour of playtime. Also the coolant seems to hit 40C whether the CPU is in the 70s or 80s, so I think its hauling the mail appropriately. I have the pump running at "Balanced." I could move it to "Extreme." I tend to keep my processor/motherboard for 7 years, and just make do with GPU upgrades. I didn't want to wear out my pump, which is why I have it set to the middle level. If that's not a concern I can move it up under heavy load to see if that helps.

Also, I've got good cable management, for the record. No air flow restrictions in the main compartment.
Maybe its your case fans but, you say your fans top out at 1400 rpm yet your cooler comes with ML fans that top out at 2400 rpm, so it begs the question what fans are you using and where? Not only 2400 rpm but those ML fans go up to 4.2 mm-H2O at 75 cfm and are considered high quality Static Pressure fans essential for radiators to perform at their best. Other than Blower style cooled GPUs which blow it's GPU heat out the back of the case, heat dump is always an issue with most aftermarket coolers and compounded when using a airflow restricted case so, for that matter what case are you using and what orientation are the fans and what GPU specifically? I'm not trying to be condescending with these questions I'm trying to help I believe we have a chance here to help you if we have a better picture of what we are looking at here. Then there is stupid shit like are you using the dust filter with the top mounted AIO? even the best of us can do stupid so be as clear as possible without those things we already know from previous post.
 
There are lots of good points made here by others however if we are going to help you a bit more info is needed.

Maybe its your case fans but, you say your fans top out at 1400 rpm yet your cooler comes with ML fans that top out at 2400 rpm, so it begs the question what fans are you using and where? Not only 2400 rpm but those ML fans go up to 4.2 mm-H2O at 75 cfm and are considered high quality Static Pressure fans essential for radiators to perform at their best. Other than Blower style cooled GPUs which blow it's GPU heat out the back of the case, heat dump is always an issue with most aftermarket coolers and compounded when using a airflow restricted case so, for that matter what case are you using and what orientation are the fans and what GPU specifically? I'm not trying to be condescending with these questions I'm trying to help I believe we have a chance here to help you if we have a better picture of what we are looking at here. Then there is stupid shit like are you using the dust filter with the top mounted AIO? even the best of us can do stupid so be as clear as possible without those things we already know from previous post.

OK, I'm that guy. It doesn't "beg the question," it raises the question. To your point though, you've correctly noted that I've traded out the fans it came with for a pair of SP120 RGBs. The case is a Cougar Panzer S, with which I'm very pleased. No dust filter on top. Took it out. The fan configuration is the radiator in the top compartment with the fans below in the main compartment pushing air out the top. I'm running an EVGA 3080 FTW3 Ultra. 3 more SP 120 fans in the front, and 1 more in the back. I just finished a 3 hours session which covered Destiny 2 and Borderlands 3, and here are max stats that were recorded by HWMonitor.

Temperature
Motherboard: 38C
CPU: 90C (I don't think this was sustained)
RAM: 50C
GPU: 71C

Power
CPU: 107W/1.43V
GPU: 294W

Most cores hit 4.6GHz, but 2 hit 4.8GHz. Similarly, most cores pulled under 10W, but 2 went to 18W, which were also the 2 cores which saw 100% utilization.

The 3 fans in the front of the case are running at max during these sessions, and the dust filter on them is clean. The room is probably 22-25C, and there's excellent air flow within the case; absolutely no cables strewn about in the main compartment. I've not tested reversing the air flow on the radiator, but this video, though several years old, indicates that air flow direction won't lead to much improvement in thermals (see 2:00).
 
I just skimmed it but that video seems like trash. They only tested each configuration for like 10 minutes which was probably not enough to heat soak the radiator where the differences would matter. Additionally if you look they are using a blower video card which would be completely different then the scenario you have with a GPU which is exhausting into the case and in the case of a 3080 FTW3 ultra a fair bit of the head actually goes straight up out the backplate (I know because i own one). This is going to lead to a lot of heat (350-400w) and a lot of it is going straight into your radiator. You don't have to be a scientist to understand that the hotter the air going into the radiator the harder its going to have to work / hotter it's going to get. Just like if it gets hotter in your room.
 
After posting all that info, I did some further checking, and I'm becoming more and more convinced that everything I know is wrong. The high end Ryzens just want to run at 90C and will boost until they hit it. Lowering the temperature is just going to allow it to boost more, but it's still going to try to hit 90C.

To borrow from Kubrik, a better title for this post might be "Dr. Hallock, or How I Learned to Quit Worrying and Love PBO."
 
Nice case I can see why you like it. Airflow shouldn't be an issue.
My recommendations:
H100i at front high position
ML fans (only) in push/intake
1 additional fan on front panel w/aio push/intake filtered
1 fan at case floor push/intake filtered
1 rear push/exhaust unfiltered
3 fans push/exhaust top of case unfiltered
remove all expansion slot covers for all unused pcie slots
pump full throttle all the time.
Start at minimum 1100-1200 rpm all fans or 800 min for SPs set curve from there work it down from there to get your noise/performance comfort level
do not trust the corsair water temp sensors
If you need the bling or temps are still not where you want them get some ML RGB fans for push/pull on front aio.
Check the 5900x vcore many motherboard bios default to high and many can get away with less.
Use Ryzen Master to test lower Vcore. For example my MH8 defaults to 1.425 and I set it to 1.35 with an OC completely stable.
Let us know if this helps!
 
That's a pretty darn high water temp and a serious concern. It won't be doing much cooling with water temp that high.
This starts to get into the temp range where I'd normally see 8c+ consumer cpus pegged from synthetic benchmarks.

Is AMD trying to close the IPC gap by keeping their cpus pegged on zen3?

OP I'd worry about fsilures when they happen.
Figure an aio goes 3 years, you got your $ out of it.

Maybe increase intake volume vs aio curve so you aren't creating a psi problem at full boost.
 
Nothing wrong with your temps. I wouldn't worry about it unless the fans are making too much noise for you.
 
Nice case I can see why you like it. Airflow shouldn't be an issue.
My recommendations:
H100i at front high position
ML fans (only) in push/intake
1 additional fan on front panel w/aio push/intake filtered
1 fan at case floor push/intake filtered
1 rear push/exhaust unfiltered
3 fans push/exhaust top of case unfiltered
remove all expansion slot covers for all unused pcie slots
pump full throttle all the time.
Start at minimum 1100-1200 rpm all fans or 800 min for SPs set curve from there work it down from there to get your noise/performance comfort level
do not trust the corsair water temp sensors
If you need the bling or temps are still not where you want them get some ML RGB fans for push/pull on front aio.
Check the 5900x vcore many motherboard bios default to high and many can get away with less.
Use Ryzen Master to test lower Vcore. For example my MH8 defaults to 1.425 and I set it to 1.35 with an OC completely stable.
Let us know if this helps!
These are goodsuggestions. It sounds like the temps are fine based on other posts (from an operational standpoint). The ML RGB fans unfortunately only go to 1600 rpm, so not a big improvement. The reorienting of the radiator to the front could likely be done without too much effort, and I may consider that as an option.

Honestly, I think the suggestion which would have the greatest impact would be manually undervolting the Vcore. I've noticed that all cores hit 1.425 as a max at one point or another. When you set the vcore to 1.35, does it treat that as a new ceiling and still allow the cores to run at lower voltages when not under load? I would think that would be the case, but just checking. That said, since I'm not going to manually OC this rig, and assuming that 90C max is just fine, is lowering the vcore just to get lower temps going to effectively lower my performance (i.e. am I just sacrifcing performance to lower my thermals unnecessarily)?
 
After posting all that info, I did some further checking, and I'm becoming more and more convinced that everything I know is wrong. The high end Ryzens just want to run at 90C and will boost until they hit it. Lowering the temperature is just going to allow it to boost more, but it's still going to try to hit 90C.

To borrow from Kubrik, a better title for this post might be "Dr. Hallock, or How I Learned to Quit Worrying and Love PBO."
No they don't want to run at 90c and 95c is the throttle point. Over 68c and you start losing boost frequency so they actually prefer lowest temps like every other piece of silicon.

You're ambient is 30c. You're temps and water temps reflect that. I think there's a disconect here...
 
Per the suggestion of owcraftsman I set the vcore offset to -0.1. I then played the Black OPs level that I know produced the higher temps which started this thread. This resulted in a max voltage of 1.394V on half the cores, and 1.387V on the others. The CPU temp maxed out at 81C, and even with less volts, 2 cores hit maxes of 4.9 Ghz, 2 at 4.8, 3 at 4.7, and the rest between 4.4 and 4.6. I wouldn't expect that they are all boosting to these levels at the same time, but it seems that with slightly less voltage, I'm still hitting the same peaks as before, but with considerably less heat.

Unfortunately I can't say how long those peaks are sustained because HWMonitor doesn't offer graphing over time on the output, and I haven't found any free software that offers this feature. No noticeable change in framerate though. I'll do some more testing over the next couple days with a range of titles before making other changes.

In the meantime, by lowering my vcore, what performance am I compromising, if any? If I'm still boosting to the same levels, is stability the only thing I really need to watch?
 
lol, that makes it as bad as I wrote the first time then.
Perhaps, but I don't think I'm going to get much lower temps without spending an obscene amount of money. If lowering the voltages by 0.1V gets me down into the 70s under load, plant the flag of victory.
 
Last edited:
I ended up landing on an undervolt with an offset of -0.125. No stability issues and the temps stay in the 65-75 range; with spikes staying below 80 so far. Why does AMD set their voltages so high by default?
 
I ended up landing on an undervolt with an offset of -0.125. No stability issues and the temps stay in the 65-75 range; with spikes staying below 80 so far. Why does AMD set their voltages so high by default?
they dont, its the board giving them too much juice.
 
they dont, its the board giving them too much juice.
+1
and keep in mind the voltage still fluctuates and ramps up and down like core clocks which is by design from AMD. Also, not all cpus can handle the lower vcore and then some more than others. (silicone lottery)
 
Negative offset reduces what the boost algorithm can push so don't make another thread going omg my chip is slow.
 
Negative offset reduces what the boost algorithm can push so don't make another thread going omg my chip is slow.
I haven't seen a reduction in clock speed as a result of the undervolt. I was getting boosts to 4.9 Ghz across the same number of cores before and after.
 
HWInfo will produce a csv file of data over time FWIW. Easy enough to pull it into Python and make some graphs.
 
To get something more concrete than just perception in game, I ran some test with Timespy Extreme. With the -0.125 undervolt, my scores for CPU, GPU, and overall were 7383, 8983, and 8700 respectively. Without it, they were 7725, 8976, and 8763. So the undervolt certainly had an impact on CPU performance, but interestingly, that impact did not translate to overall score. Oddly enough though, the temperature in both tests never got above 70C. So the games are clearly stressing the CPU significantly more than 3DMark.
 
I've got a 5950X along with the specs in my sig. All corsair peripherals with 7 QL fans and 3 ML fans on my Capellix 360 Rad. Built the system 3 weeks ago, and like u experienced the exact same symptoms. COd uses AVX I have found and temps go up almost to 80C. Scoured the internet like u looking for some solution or piece of mind. Have gone down curve optimizer, under-volting GPU, increasing fan %, etc.

At the end of the day, I realized I was obsessing way to much and not even enjoying my new rig. I decided to leave everything well enough alone and at stock. I now just enjoy and play and have faith that my hardware will work as designed. And if it breaks, oh well, replace and move on. Enjoy your new rig dude!
 
repoman0 - Thanks for the suggestion. HWInfo has been a tremendous help. I've been analyzing output from 2-3 hour gaming sessions to see things.

Island - I think we've come to the same conclusion. I got rid of the undervolt and started testing with HWInfo. I found that in Borderlands 3 I had a few peaks up to 84C, but it stayed between 70C and 80C pretty much all the time. Destiny 2 was even lower, with temps rarely going above 60C. Tonight I played CoD Cold War, and I saw peaks of 89C. It seemed to stay right around 85C most of the time though.

I still have to test Jedi: Fallen Order, Halo MCC, and Zelda: Breath of the Wild, and I'll add those findings here for posterity as I get around to it.
 
I'd like to invite both of you to joining me on the darkside and buying a D15. I have a 5900X and am doing a lot better than both of your outcomes. Blowing those Corsairs out of the water.. I can't get mine to go over 61C after 10 minutes of a Prime95 24-thread torture test blend in a 19.6 liter mITX case, holding 4.5 to 4.6GHz on all cores. In games as you're describing, I'm not going much over 50C. It does sound like temps you're getting on that AIO are within expected range.

Could be, but my original budget for this rig was $1,000 less than what it became. I sold off my 1080ti for a net return of $400, so that helped cover some of that extra cost, but I can't justify additional expenses unless there is a clear reason to do so for stability or preventing damage to parts. Neither of those appear to be in play.

By the way, I have temperature data for Jedi Fallen Order. During a 3 hour session it ran at 65C-70C.
 
Back
Top