R7 5800X3D cooling properly?

[fastfishy2]

I'm cooling my 5800X3D with a Noctua NH-D15s Chromax, in a Phanteks P300 case w/ mesh front panel replacement.

 

Temps are just fine during gaming, the worst "gaming" temp I've been able to get is 80-84c when loading a world in Minecraft Win10 with render distance at over 100 chunks, which makes extremely heavy use of the CPU while it's loading all the chunks in.

 

Cinebench R23 hovers around 76c

 

Only thing I'm worried about is OCCT Small Data Set test, because it basically goes straight to 90c and stays there, cores are usually holding a bit above 4ghz at that point, sometimes higher, but I don't really want to leave it on that long. 

 

Does this seem bad, or is it just how Le Toasty Boi 5800X3D behaves on certain benchmarks?

[freeagent]

That is really hot. 90 is brutal. My sample is very easy to cool, I can run it semi passively on my PS120SE, and FC140.

 

FWIW I score about 152xx with my X3D and temps are usually 63-65. 75-80 semi passively.

[fastfishy2]

2 minutes ago, freeagent said:

That is really hot. 90 is brutal. My sample is very easy to cool, I can run it semi passively on my PS120SE, and FC140.

 

FWIW I score about 152xx with my X3D and temps are usually 63-65. 75-80 semi passively.

yeah, as I said though the only test it gets to 90c with is OCCT small data set. Medium and large data sets it stays well below 90c, low 80's at worst but usually high 70's, and it does well enough in cinebench etc.

[Shimejii]

Minecraft puts a LOT of load on the CPU if you are hosting the server. Mine does the same behavior and i use an NH-D15 as well. Using Fan Control, i usually put it at around 70% fan speed while in most games, and you can do 100% its not louder then ingame stuff at all. Minecraft requires the faster fan speeds if its a local game.

 

If you dont like using PBO2 Tuner and doing a -25 all core, Reducing PPT to 120, TDC to 80 and EDC to 110, you can try it in the bios.

[fastfishy2]

Just now, Shimejii said:

Minecraft puts a LOT of load on the CPU if you are hosting the server. Mine does the same behavior and i use an NH-D15 as well. Using Fan Control, i usually put it at around 70% fan speed while in most games, and you can do 100% its not louder then ingame stuff at all. Minecraft requires the faster fan speeds if its a local game.

yeah I ran that scenario with minecraft specifically because I know Bedrock edition in particular can and will use most of your CPU's resources if it feels like it needs to, and for 100+ chunk render distance it starts to put heavy load on 6 out of 8 cores and a moderate load on the remaining 2. Highest power draw I've managed to get in any of my games so far.

 

Do you think the temps I outlined are relatively normal? I never get near danger temps in most benchmarks except one, and temps stay well below 80c during actual gaming, but I'm not sure. Used to have a R5 3600 so this thing is new to me.

 

[Shimejii]

Just now, fastfishy2 said:

yeah I ran that scenario with minecraft specifically because I know Bedrock edition in particular can and will use most of your CPU's resources if it feels like it needs to, and for 100+ chunk render distance it starts to put heavy load on 6 out of 8 cores and a moderate load on the remaining 2. Highest power draw I've managed to get in any of my games so far.

 

Do you think the temps I outlined are relatively normal? I never get near danger temps in most benchmarks except one, and temps stay well below 80c during actual gaming, but I'm not sure. Used to have a R5 3600 so this thing is new to me.

 

Overall id say your fine. Its just use PBO2 Tuner on windows, or if you have a BIOS that does it for you do that. 

[fastfishy2]

3 minutes ago, Shimejii said:

Overall id say your fine. Its just use PBO2 Tuner on windows, or if you have a BIOS that does it for you do that. 

oh I've heard about that program but I'm hesitant to run third party software to modify a locked CPU because I'm afraid I'll break something or void warranty

[Shimejii]

1 minute ago, fastfishy2 said:

oh I've heard about that program but I'm hesitant to run third party software to modify a locked CPU because I'm afraid I'll break something or void warranty

Its pretty easy and it wont break it. Infact it will make it run better. If you want the most stability do -20, -25 works perfectly fine for me, some people get -30 pretty easily as well. Then the 2nd menu you do the stuff i told you to do and youll see even better results.

[fastfishy2]

4 minutes ago, Shimejii said:

Its pretty easy and it wont break it. Infact it will make it run better. If you want the most stability do -20, -25 works perfectly fine for me, some people get -30 pretty easily as well. Then the 2nd menu you do the stuff i told you to do and youll see even better results.

I might hold off on that if you think my CPU temps are fine at the moment. I am curious, what temp does your 5800X3D hit on OCCT small data set and Cinebench r23? What cooler do you have?

[johnno23]

Might i ask how long the 5800x3d has been in your system. ?

temps seem a little high but if you only recently installed it might be as simple as an unequal mounting pressure.

I would lay the pc on its side so the the CPU is flat and the cooler is standing on top of it so to speak.

Then carefully loosen mounting screws. I would carefully with the flat of my hand apply a little pressure to the cooler and twist gently but not lift in any way.

Twisting the cooler to just ensure that paste is squeezed out a little. then mark the screws if possible and begin to tighten them up again ensuring that you get the same tension on each screw.

That would be my first attempt at trying to correct temps and takes around 4 or 5 minutes.

[freeagent]

11 minutes ago, fastfishy2 said:

oh I've heard about that program but I'm hesitant to run third party software to modify a locked CPU because I'm afraid I'll break something or void warranty

Have a looksey in the PBO section, and you should Curve Optimizer, from there set it to all core -30 and enjoy better boosting with lower temps. If -30 somehow gives you a problem, -25 should clear it up.

[fastfishy2]

13 minutes ago, johnno23 said:

Might i ask how long the 5800x3d has been in your system. ?

temps seem a little high but if you only recently installed it might be as simple as an unequal mounting pressure.

I would lay the pc on its side so the the CPU is flat and the cooler is standing on top of it so to speak.

Then carefully loosen mounting screws. I would carefully with the flat of my hand apply a little pressure to the cooler and twist gently but not lift in any way.

Twisting the cooler to just ensure that paste is squeezed out a little. then mark the screws if possible and begin to tighten them up again ensuring that you get the same tension on each screw.

That would be my first attempt at trying to correct temps and takes around 4 or 5 minutes.

had it for a few weeks. I was very careful to make sure the mounting screws were both done up to the same tightness and that the cooler was sitting evenly, I think I had my case lying on its side when mounting the cooler too. 

 

It's really weird though, my temps in synthetic benchmarks are at or slightly below what others are reporting for that CPU, with the exception of OCCT small data set. on Medium data set it struggles to reach 72c, Cinebench R23 sits happy at 75-76c, even that massive chunk loading stress test in Minecraft only got it to the low 80's (while loading up almost the whole CPU). It's just OCCT small data set that it really struggles with, and I don't know why. 

 

So am I just having a problem with one benchmark or do all of these temps look too high to you? OCCT is using AVX2 by the way.

[johnno23]

Your temps are in no way extreme just a tad higher than expected which is why I was leaning towards a possible issue with mounting pressure.

You could try simply using the bios to undervolt a little. there are some excellent guides on youtube and you can simply watch and take notes.

You might just be a little unlucky and have a piece of silicon pulling more power than others. More power more heat but it can change according to what apps are running and dont forget many games do not impact the CPU anywhere near as much as an apps designed to test a CPU.

I see no reason to worry but there are several ways in which to reduce the temps if it gives you a little more peace of mind.

[johnno23]

30 minutes ago, Shimejii said:

Its pretty easy and it wont break it. Infact it will make it run better. If you want the most stability do -20, -25 works perfectly fine for me, some people get -30 pretty easily as well. Then the 2nd menu you do the stuff i told you to do and youll see even better results.

agreed. I got to -30 with ease but had a little instabilty in some games so dialed back to -20 about 18 months ago and since then rock solid on a 5900x

[fastfishy2]

2 minutes ago, johnno23 said:

Your temps are in no way extreme just a tad higher than expected which is why I was leaning towards a possible issue with mounting pressure.

You could try simply using the bios to undervolt a little. there are some excellent guides on youtube and you can simply watch and take notes.

You might just be a little unlucky and have a piece of silicon pulling more power than others. More power more heat but it can change according to what apps are running and dont forget many games do not impact the CPU anywhere near as much as an apps designed to test a CPU.

I see no reason to worry but there are several ways in which to reduce the temps if it gives you a little more peace of mind.

Ok, I see. Might do a bit more digging on this. Otherwise I'll try remounting it at some stage. What temps should I be seeing on Cinebench R23 etc? 

[johnno23]

1 hour ago, fastfishy2 said:

Cinebench R23 etc

that app can have a relatively significant swing in temps on CPUs of the same type simply due to the characteristics of the silicon.

De Bauer has some decent testing showing results but in principle as long as you are within 6% - 8% of anything else of the same family it is fine. take  7 to 8% as a maximum deviation. possible swings between 3% and 6% I would expect to be pretty average. Not all CPUs are equal is the reality. 

I am unsure how you have set up your airflow but I did test my older 8700k with a dual fan noctua owl years ago and was interested how much difference the rear case fan made as the Noctua's rear fan was relatively close in a smaller case. removing the case fan had a larger impact than I expected. The test was done as the case fan was so dammed noisy and I ended up simply buying a better quality fan and the issue was solved. I was hoping to just remove it and have zero cost impact....that was not a solution LOL

 

[fastfishy2]

4 minutes ago, johnno23 said:

that app can have a relatively significant swing in temps on CPUs of the same type simply due to the characteristics of the silicon.

De Bauer has some decent testing showing results but in principle as long as you are within 6% - 8% of anything else of the same family it is fine. take  7 to 8% as a maximum deviation. possible swings between 3% and 6% I would expect to be pretty average. Not all CPUs are equal is the reality. 

I am unsure how you have set up your airflow but I did test my older 8700k with a dual fan noctua owl years ago and was interested how much difference the rear case fan made as the Noctua's rear fan was relatively close in a smaller case. removing the case fan had a larger impact than I expected. The test was done as the case fan was so dammed noisy and I ended up simply buying a better quality fan and the issue was solved. I was hoping to just remove it and have zero cost impact....that was not a solution LOL

 

Okay so, I've spent the past hour rerouting cables so that my rear case fan isn't slaved to the 140mm sitting in the middle of the NH-D15-S. I repasted, remounted, and even installed a spare 120mm noctua redux fan on the front of the heatsink (I was only using the 140mm that sits in the middle, no other front or rear fans). 

 

To put it simply, nothing changed. Remounting didn't change temperatures at all, and the extra fan on the front made perhaps 0.5-1c difference but I'm thinking that was more margin-of-error. I've removed the front fan, didn't see a definite increase in temps so it just seems that "it is what it is".

 

The overheating in OCCT SFFT & Prime95 SFFT might be to do with them being extremely cache-heavy, and the 5800X3D having a lot of it... I don't know, I'm not a processor engineer.

[fastfishy2]

1 hour ago, johnno23 said:

that app can have a relatively significant swing in temps on CPUs of the same type simply due to the characteristics of the silicon.

De Bauer has some decent testing showing results but in principle as long as you are within 6% - 8% of anything else of the same family it is fine. take  7 to 8% as a maximum deviation. possible swings between 3% and 6% I would expect to be pretty average. Not all CPUs are equal is the reality. 

I am unsure how you have set up your airflow but I did test my older 8700k with a dual fan noctua owl years ago and was interested how much difference the rear case fan made as the Noctua's rear fan was relatively close in a smaller case. removing the case fan had a larger impact than I expected. The test was done as the case fan was so dammed noisy and I ended up simply buying a better quality fan and the issue was solved. I was hoping to just remove it and have zero cost impact....that was not a solution LOL

 

I turned off all AVX instruction set options on OCCT and it could actually handle small data sets this time... it was hovering around 85-86c but going full speed on clocks and everything. It was drawing about the same if not higher package power as before but running decently cooler. 

[Mark Kaine]

5800x3D is notoriously hot. Either get better cooling or set pbo2 to -10 to -30 as already been said or just live with it, its not going to overheat or anything this is still normal temps.

 

 

Mine does 79c max during cb with pbo -30 (and with a much smaller cooler, not at max rpm) btw. good luck  

[fastfishy2]

16 minutes ago, Mark Kaine said:

5800x3D is notoriously hot. Either get better cooling or set pbo2 to -10 to -30 as already been said or just live with it, its not going to overheat or anything this is still normal temps.

 

 

Mine does 79c max during cb with pbo -30 (and with a much smaller cooler, not at max rpm) btw. good luck  

ah righto that makes sense. I've been hearing things about it being hot and AVX stress tests with small data sets seem to kick its ass. 

What cooler are you on?

[Mark Kaine]

38 minutes ago, fastfishy2 said:

ah righto that makes sense. I've been hearing things about it being hot and AVX stress tests with small data sets seem to kick its ass. 

What cooler are you on?

noctua u12s, but that shouldn't really matter... decent case cooling and pbo is what makes it run cooler... if ur scared just do -10 in pbo and see what it does, dont need to go all the way to -30... 

 

btw i also have pbo off currently,  i only see like 50-60c max during gaming etc so i dont need it, im fine, cpu/gpu around 50-60c most of the time.  but still, try it out, pbo2.

Edited June 5 by Mark Kaine
typos

[Mark Kaine]

this should cover everything you need 

 

https://github.com/PrimeO7/How-to-undervolt-AMD-RYZEN-5800X3D-Guide-with-PBO2-Tuner/blob/main/README.md

 

 

[fastfishy2]

41 minutes ago, Mark Kaine said:

noctua u12s, but that shouldn't really matter... decent case cooling and pbo is what makes it run cooler... if ur scared just do -10 in pbo and see what it does, dont need to go all the way to -30... 

 

btw i also have pbo off currently,  i only see like 50-60c max during gaming etc so i dont need it, im fine, cpu/gpu around 50-60c most of the time.  but still, try it out, pbo2.

I'll have a look if it ever gives me temperature problems during gaming, but I can't actually get it to do so in gaming loads. Thanks for the advice.