Hot Temps for 5800X3D

[3x0]

He should post a screenshot with PBO2 Tuner program with CO applied open next to HWInfo after a CB23 run. Would dispel a lot of confusion.

 

[tabascosauz]

@tabascosauz So i removed the hdd bays and replaced one of the intake fans.

Would you say its worth mounting fans on top of the rad for more exhaust too?

Just as an experiment, I put the aircon on at 18c and waited for the coolant temp to drop down significantly (23.9c in this case). Put AIO fans at 100% and pump on extreme.

Anything helps, especially airflow that actually gets to the CPU and AIO without being blocked by drive cages or GPU. If you find massive differences between ambient and motherboard temp (5-10C+), improve airflow.

• As for cooling: the results at 16C ambient don't look all that bad, considering the high Vcore and high proportion of Core power. Not the most impressive cooler.
• As for performance: clearly the scores are up to snuff and clocks are okay if not perfect. It's just that the Vcore is all over the place (way low at 90W, way high at 142W) and your lower power performance is below expected - clearly something is wonky with either BIOS, board, or CPU.

If you can't find someone to make a modded 2604 BIOS, then just do all your tweaks in PBO2 Tuner. Both CO and power. Stop doing your limits in BIOS. Even if they work, the only reason you have access to it is because Asus forgot to hide it.

You can reflash 2604 at any time if you suspect BIOS corruption, you have nothing to lose except a few minutes re-entering settings.

Really, if you aren't dealing with performance issues in games, we're just going around in circles. Just leave it be, and maybe reinstall Windows while you're at it.

 

[LordsnLadies]

I'm confused. Do you have snapshot CPU polling enabled in HWinfo?

Didn't even know about this setting until now. It was not enabled, i just used default settings.

I'd clear CMOS and stick with AGESA 1206. If that doesn't help, I'm afraid a clean OS install is in order.

I performed those tests on a win10 bootable so I could get good results on cinebench. On my win11 install the score would have been down 1000 points, which isn't accurate.

I'll try clear the CMOS and I do think a reinstall is a good idea, it's a win11 that was upgraded from win10 and one I've been using across different motherboards and ssds.

As for cooling: the results at 16C ambient don't look all that bad, considering the high Vcore and high proportion of Core power. Not the most impressive cooler

What coolers do you consider good that are AIO? This one wasn't exactly cheap, so I'm just wondering what would have been a better choice. I thought it would be better than my friend's sythe fuma rev b which costs less than half of my cooler and ofc better than the phanteks.

Really, if you aren't dealing with performance issues in games, we're just going around in circles. Just leave it be, and maybe reinstall Windows while you're at it.

This is why I have been stalling the RMA. All the games I play are able to get full clock speed even on the lower power limit settings (90/60/80 for example). Just seems like it would be more effort than its worth, because I'll be without a cpu for 1-2 weeks and there is a high possibility they'll tell me nothing is wrong with the cpu.

He should post a screenshot with PBO2 Tuner program with CO applied open next to HWInfo after a CB23 run. Would dispel a lot of confusion.

I do mention what settings I've been using. I just apply the -30 CO to all cores and those power limits in pbo tuner, nothing else. Will include PBO window in future screenshots.

 

[tabascosauz]

This is why I have been stalling the RMA. All the games I play are able to get full clock speed even on the lower power limit settings (90/60/80 for example). Just seems like it would be more effort than its worth, because I'll be without a cpu for 1-2 weeks and there is a high possibility they'll tell me nothing is wrong with the cpu.

And honestly that's reason enough not to. 5800X3D isn't designed as an all-core beast anyway. All of these minor issues including temps are to do with all-core, which games pretty much don't actually use (except for ancillary tasks like loading shaders).

As for RMA, I don't think AMD does much additional testing once they receive it, not like RMAs of GPUs or boards. As long as you can convince the rep with your RMA request, and your CPU doesn't show up with physical damage or defaced IHS, shouldn't be any problems.

I straight up wouldn't recommend an AIO. It doesn't give you any real headroom on this CPU since the real advantages of water are in all-core. Save the water for a 5900X or 5950X or Intel that you can meaningfully push to higher power. Fuma 2, Dark Rock Pro 4, U12A, D15, FC140, take your pick of high end air coolers.......there are others. Thermalright PA120 is cheap.

 

[INSTG8R]

Well like the comedy of errors that this rebuild entailed I found the last one. My case 360 ad fans are ancient 2 pin full spoed(1300rom) and run off a Molex splitter. well just like my pump in my haste I yanked the super tiny positI’ve wire oiut of the Molex. When I “repaired it” I just spliced it onto the red wire…DERP! My fans were only running at 50% sorted it out properly the other other day. Now back to average 35C idle 75C max full MT load.

 

[cattboy]

This threads a great read, appreciate everyones replies.

Consuming everything before my Kryonaut arrives to paste a new 5800x3d (side grade from 5900x cuz vCache)

Notes:

1)
AGESA 1.2.0.6 > 1.2.0.7 for best frequency?
GB only offers 1.2.0.6B, any noticeable disadvantages or advice? I pre-flashed 1.2.0.7 so will maybe downgrade after I poke around.

2)
Will manually set PPT/EDC/TDC, did this on 5900x assume same process works of setting to Mobo limits and using pbotuner2 under stress (prime95/etc) to dial it in?

Others values (for ballpark)
88ppt 60tdc 90edc
100/65/90

3)
5800x3d with 360AIO with -30CO can achieve 4450mhz at 25*c ambient is +/- 60-70* at full load prime95?

Not to derail, everyone's so active and this thread was helpful.

Hope you tame your temps LordsnLadies

cb

 

[tabascosauz]

This threads a great read, appreciate everyones replies.

Consuming everything before my Kryonaut arrives to paste a new 5800x3d (side grade from 5900x cuz vCache)

Notes:

1)
AGESA 1.2.0.6 > 1.2.0.7 for best frequency?
GB only offers 1.2.0.6B, any noticeable disadvantages or advice? I pre-flashed 1.2.0.7 so will maybe downgrade after I poke around.

2)
Will manually set PPT/EDC/TDC, did this on 5900x assume same process works of setting to Mobo limits and using pbotuner2 under stress (prime95/etc) to dial it in?

Others values (for ballpark)
88ppt 60tdc 90edc
100/65/90

3)
5800x3d with 360AIO with -30CO can achieve 4450mhz at 25*c ambient is +/- 60-70* at full load prime95?

Not to derail, everyone's so active and this thread was helpful.

Hope you tame your temps LordsnLadies

cb

1. 1207 generally not viewed positively, but you may be lucky and find less or no clock loss. GB probably have a thread for beta BIOSes out there on some forum

2. If you have the cooling for it and don't run all-core 24/7 there's no need for power limits. Temp reduction should first come from CO offset (-20/-25/-30 etc). Gaming or daily use there shouldn't be much need for lower limits. On mine I get full CB score (15000-15200) at 105/80/100 so that's what I run, because the rest of power budget is unnecessary outside of BBP/SFT ycruncher or linpack. If you're hammering it in CPU-heavy ycruncher tests you will need more power budget and cooler, but that's not a real world workload.

3. Not sure why you would consider CPU-heavy tests in Prime95 (smallest/small FFT) for Zen 2 or 3 CPU. Use R23 if you want a more realistic load or scores to compare, use ycruncher/OCCT/Linpack if you want a stress test. If you can hold 4450 all core in R23 (not hard at all with a -25 or -30 offset), you'll be able to do that in most things.

 

[QuietBob]

AGESA 1.2.0.6 > 1.2.0.7 for best frequency?

Get 1206 for maximum clocks, many owners reported lowered boost with 1207.

Will manually set PPT/EDC/TDC

As mentioned, no need for manual limits. All you need is CO, -30 should be stable for most scenarios. With adequate cooling, you'll be able to run all core loads at acceptable temperatures.

5800x3d with 360AIO with -30CO can achieve 4450mhz at 25*c ambient is +/- 60-70* at full load prime95?

There's no way you can keep a 5800X3D below 70c at 25c ambient with traditional cooling. With a 360 AIO, you'll top out at around 85c in R23 on stock , and hopefully about 75c with -30 CO given proper airflow. If you can keep the CPU below that temperature, you'll see the maximum 4450 boost across all cores.

 

[cattboy]

Thanks for the replies, got the beast up today.

Need to fiddle with RAM more and find out how to get 4550mhz constantly while gaming, but so far I'm happy.

Edit* 1206b
Default ppt tdc edc
Latest chipset 4.11.xxxxx

 

[Johnthedon1]

90c is fine for a stress test ur never going to do anything on ur pc that will push it that far,
You could use pbo optimiser with a negative offset to try help

 

[iGigaFlop2]

I was wondering about temps also because it runs a good bit hotter than my tuned 5950x, i just got it up and running about 10 min ago its on a kraken x72 cooler with fans and pump running .

 

[eidairaman1]

I was wondering about temps also because it runs a good bit hotter than my tuned 5950x, i just got it up and running about 10 min ago its on a kraken x72 cooler with fans and pump running .

3D cache is why

 

[Tropick]

I was wondering about temps also because it runs a good bit hotter than my tuned 5950x, i just got it up and running about 10 min ago its on a kraken x72 cooler with fans and pump running .

I just got finished building a custom watercooling loop with 420mm and 360mm rads and even with the pump running 75% speed it tops out at 73C during stress tests. The 5800X3D is just a hot running chip unfortunately.

 

[freeagent]

I think the opposite. For me it is such a cool running chip I could run my cpu cooler with no fans on it, and just the two stock fractal fans up front and see no signs of remotely overheating while running Linpack, OCCT, Cinebench, F@H, WCG, etc..

 

[Tropick]

I think the opposite. For me it is such a cool running chip I could run my cpu cooler with no fans on it, and just the two stock fractal fans up front and see no signs of remotely overheating while running Linpack, OCCT, Cinebench, F@H, WCG, etc..

Seriously? Are you using a Ryzen offset bracket? I've repasted twice now and still can't get it below the low 70s at its peak, It's baffling. Every time I think I've figured out a way to tame it I get nearly the same numbers. Never had it overheat but mid/high 70s for a relatively low clocked 8-core CPU on a custom WC loop just seems too high.

Edit: I know it's not a problem with loop flow because my heavily, heavily overclocked 6950XT averages 36c during Time Spy stress test and hotspot maxes out at 43c during the same.

 

[freeagent]

Seriously? Are you using a Ryzen offset bracket? I've repasted twice now and still can't get it below the low 70s at its peak, It's baffling. Every time I think I've figured out a way to tame it I get nearly the same numbers. Never had it overheat but mid/high 70s for a relatively low clocked 8-core CPU on a custom WC loop just seems too high.

Edit: I know it's not a problem with loop flow because my heavily, heavily overclocked 6950XT averages 36c during Time Spy stress test and hotspot maxes out at 43c during the same.

Yeah it’s pretty crazy. And no one believes me lol. I took a few screens and you can see the cores and ccd are nicely grouped together tightly. When I run a fan the temps are a little more spread apart. The temp in my space is 18-19c right now, the furnace comes on to keep us from dieing and then turns off

 

[cattboy]

For me it is such a cool running chip

What cooler?
What thermal paste?
Did you apply paste in X format or evenly spread?
Curious how you got yours so low as mine always peaks to 80c and my MC is not close to 15k on 360AIO, TG kryo spread, on 1207agesa

 

[freeagent]

What cooler?
What thermal paste?
Did you apply paste in X format or evenly spread?
Curious how you got yours so low as mine always peaks to 80c and my MC is not close to 15k on 360AIO, TG kryo spread, on 1207agesa

Thermalright FC140
Thermalright TF7
I use the spatula
Not sure?

 

[cherryberry]

Make sure to take the plastic off lol.

 

[Space Lynx]

Sounds a bit warm as mine will run a 10GB load in Linpack 30 times at 4400MHz and barely crack 80 with ARGB fans.

Optimum Tech over on YouTube has already demonstrated that 5800x3d's vary wildly in temps.

His latest video about his new 4090 PC talks about it some.

 

[freeagent]

Optimum Tech over on YouTube has already demonstrated that 5800x3d's vary wildly in temps.

His latest video about his new 4090 PC talks about it some.

I made that post while running PA120.

 

[candymancan21]

My 5800x 3d on ny custom loop with a gpu block on my 6900xt and two reservoirs and two 360mm fans. With 11 thermaltake fans in the case and on the rads.. I get 84c during prime 95 small ffts.

But this is using conductonaught liquid metal. When i use arctic silver 5 and MX4 i was getting instant 90c.

However i noticed last night when gaming playing warhammer total war 3 and watcjing youtube on my 2nd monitor. My temos did peak at 86c on the CCO and 81 for the rest. But average was in the 60-70s. The peaks were interesting though. And made wonder if something happened to the liquid metal over the last 3-4 days. My idle temps stayed at 43c as well 30 min after gaming.. prior they were 36-39 depending on room temp.. so i gatta keep an eye on this.. It coulda just been a hotter room from the stove .. Or it might be that i have 4 sticks of memory now. And my flck is at 1867 and memory is at 3733.. not sure if the increase in flck and 4 sticks of ram vs 2 played a roll in the increase in temps. OR maybe it was me watching youtube while gaming on two monitors.. maybe increasing cpu usage and thus increasing temps.. i think it was probably me watching youtube when also gaming that did it.

Interesting because prior to liquid metal i didnt see that gaming but i didnt watch videos.. But my room swings in temps due to the wood stove in the house again..

Cinebench23.. prior to liquid metal i was getting high 70s like 76-78..... after liquid metal i got low 70s for an average 5c drop in temperature. Cores averaged 69-71c.

Idle temps also dropped from 39c to 36c.

Im using an older EK Supremacy Evo block btw. Liquid metal helped alot.. i ran lprime95 small ffts for 4 hours and max temos didnt go above 84. Again prior INSTANTLY it went to 90.


I also did swap orientation of my block i dunno if that played a role. Now the evo inlet is on the left and outlet on the right. But i dont think that matters as the jet puts water in the middle anyway.

 

[Miroo/\:o:/\]

Hi,
I am jumping in this topic as I am experiencing thermal issues that are badly affecting the CPU performances.
I recently upgraded to a 5800X3D ( from intel platform...so I am a total ignorant of AMD CPU/BIOS/Motherborad)

The problem is that beast is idling at Core Temp of ~50°c which obviously lead to thermal limitations under load in CB R23 ....it peaks straight away to 90°c and then start reducing freq to 4.0~4.15 GHZ
I read the entire topic and tried CO in bios and in PBO2 105/65/90 curve -30. What ever I do it can't handle more than Vcore 1.1v sometimes even lower. I did repaste re-seat and get some improvements but very little.
So I am wondering if the issue is related to my AIO ....may be it isn't enough to handle this heat-monster .

This is a CB R23 test with stock settings ....score 12454pts

View attachment 278334

CB R23 CO(BIOS) @-30...... score 13164

View attachment 278336

CB R23 test with CO (BIOS) @-30 all cores ...score13164 pts




It goes better but still far from optimal performances of the CPU

 

[3x0]

So I am wondering if the issue is related to my AIO ....may be it isn't enough to handle this heat-monster .

Are you sure the pump is working/connected?

 

[Miroo/\:o:/\]

Are you sure the pump is working/connected?

Yes it is running ...at load it runs @2900ish RPM...monitored from CAM



Also Sometimes at cold starts I hear weird sound from the pump like a bad cluntch