Question 5800X3D running hot

[Jimbob1]

I've been playing DayZ in 4K, and I'm getting temperatures maxing out at around 86 degrees. When I run R23 I'm getting temperatures at 90-92 degrees. Would you say these temperatures are too hot or that I'm okay. I am getting very concerned on the temperatures when gaming, and wondering if anyone could help.

I've got the Corsair H115i. I'm not sure what I should do

 

[Phaaze88]

So what would you suggest I do? I’ve tried reapplying the paste. Exactly the same temperatures. I’m running Cinebench as we speak and I’m hitting 90 degrees 3 minutes in.

Yeah, paste is just a bridge - one of 3 layers connecting the cpu die(s) to the cooler's cold plate. If you applied paste several times and nothing changed, you have to look elsewhere - like what's being suggested here:

Hot Temps for 5800X3D

hello guys I recently upgraded my 5600X to a 5800X3D. My understanding is that this chip due to the v-cache design runs a bit hotter than a 5800X per say. Now I won't compare it to my 5600X as that was a Ryzen 5, this is a Ryzen 7, so more cores, etc. CPU is the only thing that changed...

www.techpowerup.com

See posts,
#10: "If you want to bring down the temps (and see if the contact is the problem), use PBO2 tuner to do your Curve Optimizer and set some lower power limits. eg. all-core -30 and ~100W should cut temps by 10-20C, if contact is not a problem. You shouldn't need more than 120W to achieve max all core performance at 4450MHz."

#18, with PBO2 settings in a snapshot that you can try...

#21: "You can adjust power to your liking, just keep in mind that all-core clocks stop scaling at 4450MHz. Only BCLK can go further. So adjust power as necessary to max out at 4450MHz, and make sure there's no major clock stretching (discrepancy between Core Clock and Effective Clock during test) or performance loss.
It's best to find a combination of PPT/TDC/EDC that works well. Adjusting PPT alone doesn't always work too well for actual performance."



TL;DR: Doing -30 on all cores in PBO2 and nothing else should see improvements.

 

[zx128k]

Be careful undervoltaging the cpu as it can cause BSOD or crashes. Or at least the paranoia of problems everytime a game crashes. If the cooler is installed right and the cooling is decent enough. Then undervolting could well help. Start with small values first. Just because someone gets -25mV won't mean it will work for everyone.

 

[Karadjgne]

Yes. Because you aren't seeing 4.45GHz on anything but single core.

Ryzen isn't Intel and doesn't behave like Intel, but ppl keep assuming they are the same, ingrained thought processes from 10+ years of core-i domination over FX.

Intels don't scale with temps. They give max performance locked in until they reach their thermal cap, then throttle way back. They totally rely on the user to provide sufficient cooling to prevent that.

Ryzen are the opposite. They scale with temps, starting @ 60°C they'll knock 50MHz or so from 1 thread, then hit several threads, at @ 80°C they'll be hitting most if not all the threads down 100MHz or so etc. It's a very gradual decline that's really hard to notice without a full core burn with static load.

As for temps, as said, the stacked cache is going to bump that up some, but not all that much. What's going to affect temps the most is ambient temps, both outside and inside the case as one affects the other. It's physically impossible to cool an object below ambient temps by mechanical means, regardless of the cooler. An aio is mechanical. It uses a fan blowing across a fin that dissipates heat energy. You'd need a chemical process, such as peltier or LN2 or phase change to drop below ambient.

So whatever the outside ambient is, that's the starting point. From there add up the inside case energy, and that'll be the minimum temp you'll see at the cpu.

Cpus generally run @ 10°C average higher than external ambient. Cases with lousy airflow can hit 15-20°C, cases with excellent airflow can hit 1-6°C. Gpu should idle roughly the same as cpu.

On a 90ish watt load, 35°-40° is very normal coolant temp on an aio as it'll be case ambient temps + wattage leech - heat exchange variable.

Unless you can physically feel a difference in the input and output hoses, your aio is working as it should. There's generally @ 2° difference, which you can't feel by touch, so they should feel the same.

With that kind of cpu temp, aio pump should be 90-100% and fans will be north of 70%. If your curves/software has them running any lower than that, there's half the issue.

Personally I'd unplug the rear exhaust fan. With a top aio, it's essentially useless. It's doing 2x things not in your favor. It's adding to the local area vacuum which makes for a lower area pressure, so a good majority of airflow is headed directly for that corner. That puts a good chunk of gpu heat right at that rear aio fan, and a lot less at the front fan. Creates uneven dissipation, the rear is warmer, so less effective, the front is cooler, and more effective, but has less to work with.

Secondly, it's stealing air from the back aio fan, making matters worse as there's more local heat but less air to move it.

What works better is a chimney. The aio will have a viable source of external air from the open fan vent as well as a more even split of gpu exhaust, since all that heat is rising straight up, not being dragged sideways by an uncompromised fan. Air moving in a single direction helps airflow and keeps case ambient temps lower as its a stronger overall movement.

 

[Darkbreeze]

How long have you had that H115i in service?

Because it sounds like either your water block has gotten loose, or the paste has dried up, or the pump is getting weak, or the small orifices in the AIO pump and radiator are clogging up with organic growth. Or, you may have an air bubble trapped at the pump/water block.

What case do you have?

How many case fans, where are they and which way is each of them blowing?

Where is the radiator mounted? Which way is it configured, as intake or exhaust (in/out)?

What are the rest of your FULL hardware specs?

 

[Jimbob1]

How long have you had that H115i in service?

Because it sounds like either your water block has gotten loose, or the paste has dried up, or the pump is getting weak, or the small orifices in the AIO pump and radiator are clogging up with organic growth. Or, you may have an air bubble trapped at the pump/water block.

What case do you have?

How many case fans, where are they and which way is each of them blowing?

Where is the radiator mounted? Which way is it configured, as intake or exhaust (in/out)?

What are the rest of your FULL hardware specs?

I've recently switched from the 3700X in the past 5 days, and never had no were near the temperatures but I've also researched that the 5800X3D runs very hot, and AMD have stated that anything below 90 degrees is okay. I have two 240m fans on the radiator mounted at the top, with one 240m fan at the back and two 240m at the front. Full hardware specs can be found in my signature.

 

[BogdanH]

I've recently switched from the 3700X in the past 5 days, and never had no were near the temperatures but I've also researched that the 5800X3D runs very hot...

It's important to understand why 5800X3D (Tmax=90°C) "runs hotter" than 3700X (Tmax=95°C). What makes a difference is power consumption: 105W for 5800X3D and 65W for 3700X. That is, it's much easier to keep 3700X cool, as that's the case for 5800X3D.
So, what can be done? By assuming, that you mounted your existing cooler properly and that pump is forking fine, then the only solution would be a "better" cooler. Still, no matter what cooler, you won't be able to get 5800X3D as cool as 3700X. The reason for that is area of heat dissipation (hot spot area): 5800X3D and 3700X have the same area, but amount of heat is almost double for 5800X3D (on that same area).
To sumarize.. I think you should get 5800X3D to work at about 85°C (at full load). The thing is, at temperature close to 90°C, CPU is allready throttling (slows down) -means, you run it hot, still you don't have full performance.

Bogdan

 

[Jimbob1]

It's important to understand why 5800X3D (Tmax=90°C) "runs hotter" than 3700X (Tmax=95°C). What makes a difference is power consumption: 105W for 5800X3D and 65W for 3700X. That is, it's much easier to keep 3700X cool, as that's the case for 5800X3D.
So, what can be done? By assuming, that you mounted your existing cooler properly and that pump is forking fine, then the only solution would be a "better" cooler. Still, no matter what cooler, you won't be able to get 5800X3D as cool as 3700X. The reason for that is area of heat dissipation (hot spot area): 5800X3D and 3700X have the same area, but amount of heat is almost double for 5800X3D (on that same area).
To sumarize.. I think you should get 5800X3D to work at about 85°C (at full load). The thing is, at temperature close to 90°C, CPU is allready throttling (slows down) -means, you run it hot, still you don't have full performance.

Bogdan

When it’s hitting 90 degrees the CPU is still running at 4.3GHZ. I went into the motherboard and turned off the boost, running at 3.4GHZ and temps don’t go past 60 degrees under full load but I’m losing a lot of performance.

 

[Roland Of Gilead]

As others have said these chips can run hot. The vanilla 5800x draw up to 120w at highest boosts. The 5800x3d on the other hand draw up to 140w limited to 4.5ghz. The extra cache doesn't dissipate the heat as good as the regular 5800x though.

Running with boost off is not the answer.

Re-check cooler seating, and thermal paste. What paste have you used?

edit: are both pipes on the AIO the same temp, or roughly the same?

 

[Jimbob1]

As others have said these chips can run hot. The vanilla 5800x draw up to 120w at highest boosts. The 5800x3d on the other hand draw up to 140w limited to 4.5ghz. The extra cache doesn't dissipate the heat as good as the regular 5800x though.

Running with boost off is not the answer.

Re-check cooler seating, and thermal paste. What paste have you used?

edit: are both pipes on the AIO the same temp, or roughly the same?

SYY Thermal Paste, 2g is the thermal I've used. I've remounted it twice. I can't give a temperature for the pipes individually but the pump temperature is normally 39-40 degrees when under load.

 

[kerberos_20]

SYY Thermal Paste, 2g is the thermal I've used. I've remounted it twice. I can't give a temperature for the pipes individually but the pump temperature is normally 39-40 degrees when under load.

pump speed run at max, does hot air comming out of radiator? at 90C CPU, temperature from rad should be quite warm, you can ignore fan speeds on rad if theres no decent cold air intake

 

[Roland Of Gilead]

SYY Thermal Paste, 2g is the thermal I've used. I've remounted it twice. I can't give a temperature for the pipes individually but the pump temperature is normally 39-40 degrees when under load.

That seems like a good paste. High thermal conductivity @ 15w/mk. Is it spread thinly? How did you apply the cooler? Should be corner to corner, and then the opposites to ensure it spreads evenly. Although your temps hint at good application.

 

[Darkbreeze]

Can you post a pic of your AIO installation with the side panel off?

How to post images on the Tom's hardware forum​

 

[CelicaGT]

I'll weigh in here since I have one of these bad boys enroute from Amazon (Price is good right now, get em while you can). My research indicates that part of the issue is that the stacked cache is part of the problem, it slows heat transfer from the compute cores to the IHS. The other potential issue is the CCX layout. It's been reported that large, 8mm heat pipes work best with a NORTH SOUTH orientation if using a tower cooler, this makes sure that all the heat pipes are making contact with the singe CCX. I'm going to try and cool this piece of sh!t with a 212 EVO, apparently it can be done (my case airflow is peerless). As for your temps and clocks? They look a little high, most people report just below 90C and an R23 score of mid 14k on a 5 pipe tower, doesn't seem like the AIO's do much better due to the heat transfer issues I mentioned. Good luck! (and to me as well.....).

 

[MrN1ce9uy]

I just installed my new 5800X3D a couple hours ago, replacing 3700X. My temps seem to be higher with a typical gaming workload. I'm using a Kraken X62. I can't imagine your H115i giving you 90C....

I just installed Cinebench R23 and let it run a couple mins (it seems to have changed from years ago when it only ran once), but I let it run through about 4 times and my CPU temps reach 85C with the default silent profile on my Kraken X62. Ambient temp is 68F.

This CPU does run warmer than I expected/hoped...

What is your ambient temp?

 

[Darkbreeze]

Same; I just got mine yesterday and I went from the 5600x to the 5800x3d and it seems that I gave up stability for power. I seem to get dips in the games I play and it also the same dip. Going from about 4.1 ghz to 3.4 ghz. Driving me crazy why.

Heat. That's why. This is textbook throttling behavior. BUT, you need to start your own thread if you are looking to have a conversation or receive help. We tend to frown on thread hijacking, even when it's relevant. It's noted that you are already aware of that and are conscientious of the fact. I'm sure any of us would be glad to try and help you out if you start a thread on it.

 

[Raz_5_]

Heat. That's why. This is textbook throttling behavior. BUT, you need to start your own thread if you are looking to have a conversation or receive help. We tend to frown on thread hijacking, even when it's relevant. It's noted that you are already aware of that and are conscientious of the fact. I'm sure any of us would be glad to try and help you out if you start a thread on it.

I'll clean it up

 

[Jimbob1]

I just installed my new 5800X3D a couple hours ago, replacing 3700X. My temps seem to be higher with a typical gaming workload. I'm using a Kraken X62. I can't imagine your H115i giving you 90C....

I just installed Cinebench R23 and let it run a couple mins (it seems to have changed from years ago when it only ran once), but I let it run through about 4 times and my CPU temps reach 85C with the default silent profile on my Kraken X62. Ambient temp is 68F.

This CPU does run warmer than I expected/hoped...

What is your ambient temp?

I think the ambient temp took a big part as now I only get from 83-85 degrees max but when I play Battlefield 2042 in 4K, ultra preset it's putting a lot of stress on my CPU and getting around 81-83 degrees.

 

[Phaaze88]

it's putting a lot of stress on my CPU

Define 'a lot of stress on cpu'. What is the highest package power reading in this game - which won't be consistent, because these types of games are 'different' with every match.
Compare that with package power during Cinebench R23.

 

[Jimbob1]

Define 'a lot of stress on cpu'. What is the highest package power reading in this game - which won't be consistent, because these types of games are 'different' with every match.
Compare that with package power during Cinebench R23.

88W and 4.325Ghz.

 

[Phaaze88]

88W and 4.325Ghz.

88w is not stressing for a 140w-ish cpu and should be a cakewalk for that cooler.

Cpu cooler is in the usual top exhaust? Then the heat energy from everything below it is passing through as well, not just the cpu.
During Cinebench, not much else is doing anything besides the cpu, but the cpu is using more than 88w.
There's the way the roof of the Evolv X is designed: probably delaying heat removal. I mean, fan moving air through the rad while not sounding like a jet engine, and then the air has to turn to go out a couple side vents, can't be easy...
Combine that with the worse thermal transfer created(Z-height) by the V-cache, and yeah, you're seeing cpu thermals during games up there with Cinebench.

 

[Jimbob1]

88w is not stressing for a 140w-ish cpu and should be a cakewalk for that cooler.

Cpu cooler is in the usual top exhaust? Then the heat energy from everything below it is passing through as well, not just the cpu.
During Cinebench, not much else is doing anything besides the cpu, but the cpu is using more than 88w.
There's the way the roof of the Evolv X is designed: probably delaying heat removal. I mean, fan moving air through the rad while not sounding like a jet engine, and then the air has to turn to go out a couple side vents, can't be easy...
Combine that with the worse thermal transfer created(Z-height) by the V-cache, and yeah, you're seeing cpu thermals during games up there with Cinebench.

So what would you suggest I do? I’ve tried reapplying the paste. Exactly the same temperatures. I’m running Cinebench as we speak and I’m hitting 90 degrees 3 minutes in.

 

[MrN1ce9uy]

So what would you suggest I do? I’ve tried reapplying the paste. Exactly the same temperatures. I’m running Cinebench as we speak and I’m hitting 90 degrees 3 minutes in.

I think your temps are normal. I just ran R23 for 3 minutes exactly and I hit 90C right before 3 mins... looks like this CPU does run up to around 140W.

 

[zx128k]

high temps are a known problem on the 5800x3d, example

The last post were one person doesn't have the issue is important.

 

[Jimbob1]

high temps are a known problem on the 5800x3d, example

The last post were one person doesn't have the issue is important.

I’m confused on how his temperatures are so low. Maybe it’s an issue with the CPU I received? It’s things like that what get me paranoid and worried to why mine isn’t getting the same result.