Question Custom Liquid with Ryzen 3600x over 80°C .. is correct?

[GiovaFava]

Dears,
first, my apologize for my english but I write from Italy. I'll do my best.
This is my first post in the forum and I hope I can get your support.

I'll briefly explain my problem.
I just finished my new build for photography (Capture one) and occasional 3D design (Fusion 360) and rendering (Blender.. in training). Of course some stress test was to be done..
I would ask you if, with this set, those temperatures are to be expected or there is something wrong. (to mee they looks too high).

CPU idle temp is around 35°C.
These days room temperature is around 26-28 °C

Using softwares like PassMark or Cinebench R20, despite the custom loop (see below components), the CPU shows temperature spikes at over 72-75°C and just over 80°C when I run Blender 2.8 render processor. Blender render processor can work both with CPU only, and CPU+GPU (and GPU alone, but I don't care much) but the CPU temperature is not affected by swapping between these methods.

When the GPU is running 100%, 56-57°C is more or less its operating temperature, which looks pretty decent, without downgrading performances even if the water block is not fullcover and VRM and VRAM are passively cooled with some heatsink. (no dedicated fan)

Said temperature are not notably affected if the MSI BIOS is at standard settings, where Ryzen looks to throttle down to 4.1 GHz, or in OC Genie 4, that sets all 6 cores to 4.25 GHz. Multithread performances are instead improving by this option.
Moreover, with OC Genie HWinfo report 95W on the CPU cores; without OC Genie, the energy consumption stays around 82-85W.
I also changed the first thermal compound for the CPU from standard Master Gel to the "Maker" version, but I noticed almost no changes (maybe 1-2 degrees less).

The loop is Tank-CPU-GPU-Radiator-Tank. Radiator fans are in PUSH.

Last but not least, I limited the pump to the 4th out of 5 power level because at full throttle it vibrates a bit and makes noise. In any case the pump speed is at 3800-3900 rpm, at 100% PWM.

Components
Motherboard : MSI X570 - A PRO (updated to latest bios just yesterday)
CPU: Ryzen 5 3600x
RAM: Vengeance DDR4 2x8GB 3000MHz (XMP enabled in the BIOS)
GPU: referece GTX1060 (coming from previous rig)
COOLING: Custom water loop kit from Alphacool- Eissturm Blizzard 280, 45mm thick, 2x140mm be-quiet silent wing 3 (1200RPM), pump VPP755, dual slot 5.25 bay reservoir,
CPU water block: Alphacool Eisblock XPX
GPU water block: generic Alphacool for GPU
Thermal compound: CPU - Coolermaster master gel maker (11W/mK), GPU - Coolermaster Mastergel (std, >1.85 W/mK)
CASE: Coolermaster Master Case 600P with additional 3x 140mm standard (poor quality) fans (1200RPM)


Thanks everyone will answer!

 

[logainofhades]

Make sure you have not over-tightened the CPU block. On my AIO, I did, and saw horrible temps, until I back the screws off some.

 

[rubix_1011]

First question: Is the CPU overclocked at all?

Outside of this, have you confirmed the CPU block has seated correctly and thoroughly by looking at the thermal paste contact pattern on the block and CPU IHS if you remove the block?

Does the CPU slowly climb to 80C or does it immediately go from 35C at idle to 80C at load?

Does the CPU temp drop from 80C back to idle temps when load ceases, or does it remain?

At any point does the GPU go above 56C? What is the GPU at idle?

I'm trying to determine if there are more things in play, but the above questions will help determine that.

 

[GiovaFava]

First question: Is the CPU overclocked at all?

Outside of this, have you confirmed the CPU block has seated correctly and thoroughly by looking at the thermal paste contact pattern on the block and CPU IHS if you remove the block?

Does the CPU slowly climb to 80C or does it immediately go from 35C at idle to 80C at load?

Does the CPU temp drop from 80C back to idle temps when load ceases, or does it remain?

At any point does the GPU go above 56C? What is the GPU at idle?

I'm trying to determine if there are more things in play, but the above questions will help determine that.

I did not detached that waterblock to see if the paste is evenly spread... Should l? I am not damaging the joint if I remove the waterblock before screwing?
I used the "rice grain" approach.

I'll firstly check if slightly untightening the screws something change, even if the load is actually due to springs, more than the screws themselves.

The OC status is the "automatic" one done by the mobo. The CPU anyway touched 80 °C even with just turbo boost enabled. Yet...yes, now it's OC.

Many thanks! I'll let you know if something change!!


Inviato dal mio HUAWEI VNS-L31 utilizzando Tapatalk

 

[rubix_1011]

I did not detached that waterblock to see if the paste is evenly spread... Should l? I am not damaging the joint if I remove the waterblock before screwing?
I used the "rice grain" approach.

You can, it won't damage anything. Worst case scenario you just need to reapply thermal paste before re-seating the block.

I'll firstly check if slightly untightening the screws something change, even if the load is actually due to springs, more than the screws themselves.

Depending on whether they are actually secured all the way or only part way, you might need to secure them more. With spring tensioned CPU blocks, I have found that the springs should manage the correct hydraulic pressure to be applied, assuming you are applying adequate torque and threading to the mounting screws.

The OC status is the "automatic" one done by the mobo. The CPU anyway touched 80 °C even with just turbo boost enabled. Yet...yes, now it's OC.

This is likely your culprit. I would reset this to factory settings, then test again. Without knowing what your system is doing without modifying clock speeds and voltages, it is difficult to gauge whether the changes the overclock is making is the cause, or if you have problems with he cooling system in any way. Many times, automatic overclocking settings overvolt for stability but adding voltage per clock cycle means higher temps.

 

[GiovaFava]

Hello!
Tonight I did some more trials, both as default an with automatic overclock.
Probably the lower room temperature compared with last days, helped in having maximum 70°C, that maybe does not represent a big deal for the CPU.
Yet is worth you be aware that to cool down from load removal it took several seconds (… around 20 to be back to idle).
I don't know if I can... I hope yes. I attache here some charts hosted in OneDrive.
https://1drv.ms/b/s!AsR_fXUehjYOkoAucZjmAw57HN3qNw?e=h7bZsc

Looking at the charts… slow cooling ramp, strange secondary peaks at 0 load… Can it be that my Ryzen has some failure in internal dissipation, under the lid??
I have 2 days left to send it back to Am..n ..should I?

Inviato dal mio HUAWEI VNS-L31 utilizzando Tapatalk

 

[GiovaFava]

You can, it won't damage anything. Worst case scenario you just need to reapply thermal paste before re-seating the block.

Here it is... To me it looks evenly distributed.

Anyway I am packing it to return

Inviato dal mio HUAWEI VNS-L31 utilizzando Tapatalk

 

[jon96789]

TBH, I think your temps are fine... Several reviews have shown that the latest Ryzen CPUs run a lot hotter than expected constantly running at higher voltages. I have a 3900X and the CPU idles at about 38 degrees. When using Prime95 or Cinebench 20, the CPU temps peaks at 90 degrees but usually averages about 80-85 degrees (I have a Corsair H115i RGB Platinum AIO cooler). The voltages on my CPU is always at about 1.48 volts which is extremely high.

When I had the stock AMD Wraith Prism air cooler, the CPU would peak at 95-98 degrees.

What you do have to keep an eye on is the MSi 570A Pro motherboard. This board (and the other lower end MSi boards) has probably the poorest and cheapest designed voltage regulator design, using a 4+2 design (some other brand boards have 14+2 designs). If you try to overclock the CPU, the regulators will overheat and cause the CPU to thermal throttle (i.e. slow down to a lower clock speed)

 

[GiovaFava]

TBH, I think your temps are fine... Several reviews have shown that the latest Ryzen CPUs run a lot hotter than expected constantly running at higher voltages. I have a 3900X and the CPU idles at about 38 degrees. When using Prime95 or Cinebench 20, the CPU temps peaks at 90 degrees but usually averages about 80-85 degrees (I have a Corsair H115i RGB Platinum AIO cooler). The voltages on my CPU is always at about 1.48 volts which is extremely high.

When I had the stock AMD Wraith Prism air cooler, the CPU would peak at 95-98 degrees.

What you do have to keep an eye on is the MSi 570A Pro motherboard. This board (and the other lower end MSi boards) has probably the poorest and cheapest designed voltage regulator design, using a 4+2 design (some other brand boards have 14+2 designs). If you try to overclock the CPU, the regulators will overheat and cause the CPU to thermal throttle (i.e. slow down to a lower clock speed)

Hi!
About the MSI, the vrm design should be 8+2... I do know that better card on the market have more phases but I read pretty decent review, looking value for money. I am non interested in super overclock or in future upgrade to 3950x or whatever.

Consider that my 3600x with the (horrible, shameful) stock cooler on default setting was running at 86-87 °C throttled down to 3.9 GHz (!!!) on all 6 cores.
Honestly speaking, I read about 60-65 under load.

What do you think about 20-25s to cool down from load removal? (See the pic)


Now...I can understand that AMD released a cheap cooler that does not handle overclocking but at least the standard turbo boost....

Anyway... It is gone. Next week I'll have a new one and I'll see. In case something change I'll tell you!

Many thanks!

Inviato dal mio HUAWEI VNS-L31 utilizzando Tapatalk