Question 5800x high temps in custom loop

[DadduDavid]

Hello, I am seeing some unusual temps for my 5800x. I have a custom loop installed recently, covering both the primary GPU and CPU.

Specs:

CPU - R7 5800x with bykski CPU-XPR-C-M waterblock
GPU1 - Gigabyte RTX 3080 Gaming OC 10G with bykski N-GV3090GMOC-X waterblock
GPU2- Colorful GT 1030 2G LP (Needed as I have 6 monitors), GPU fan unplugged as it's noisy.
Motherboard: Gigabyte AORUS B550M PRO AX
360 copper rad with 3x Arctic P12 ARGB fans as intake
240 copper rad with 2x Jonsbo FR-901 ARGB fans as exhaust
The pump is set to 80% at all times.


CPU idling at 47-50c, spikes to 70c then stabilizes at 51c when I start watching Youtube.
When running Cinebench R23 multi-core, the CPU almost instantly hits 90c and throttles (Both HWInfo64 and Ryzen Master).
I don't believe the temperatures are normal for a custom loop. I have pasted the CPU block once with random brand thermal paste, checked the spread and it was good. Then I wiped all of them and applied quite a fair amount of Kryonaut Extreme. Seems unlikely to be poor contact between the water block and CPU to me.


Curve optimizer all cores -30, PBO limits disabled
3080 idles at 42c, stabilizes at 62c when running FurMark. The loop should be fine.

PC image, BIOS settings, and Hwinfo sensors when running cinebench

Any help would be appreciated!

 

[drea.drechsler]

Hello, I am seeing some unusual temps for my 5800x. I have a custom loop installed recently, covering both the primary GPU and CPU.

Specs:

CPU - R7 5800x with bykski CPU-XPR-C-M waterblock
GPU1 - Gigabyte RTX 3080 Gaming OC 10G with bykski N-GV3090GMOC-X waterblock
GPU2- Colorful GT 1030 2G LP (Needed as I have 6 monitors), GPU fan unplugged as it's noisy.
Motherboard: Gigabyte AORUS B550M PRO AX
360 copper rad with 3x Arctic P12 ARGB fans as intake
240 copper rad with 2x Jonsbo FR-901 ARGB fans as exhaust
The pump is set to 80% at all times.


CPU idling at 47-50c, spikes to 70c then stabilizes at 51c when I start watching Youtube.
When running Cinebench R23 multi-core, the CPU almost instantly hits 90c and throttles (Both HWInfo64 and Ryzen Master).
I don't believe the temperatures are normal for a custom loop. I have pasted the CPU block once with random brand thermal paste, checked the spread and it was good. Then I wiped all of them and applied quite a fair amount of Kryonaut Extreme. Seems unlikely to be poor contact between the water block and CPU to me.


Curve optimizer all cores -30, PBO limits disabled
3080 idles at 42c, stabilizes at 62c when running FurMark. The loop should be fine.

PC image, BIOS settings, and Hwinfo sensors when running cinebench

Any help would be appreciated!

Reported temp spikes like that used to be fairly common with Zen 2 before changes to temp reporting. It's an effect of the tiny 7nm geometry, hot-spot temp sensors and boosting algorithm that lifts individual cores to max clocks from idle for extremely short duration. No amount of cooling, short of sub-ambient, could tame the spikes. At some point after the release of Zen 3, I believe, AMD "fixed" it with updated AGESA and possibly chipset drivers. At least that's what I saw with my motherboard after BIOS and chipset driver updates. I don't suppose that exactly fixed the localized temp spiking during boosts since that's inevitable with high power dissipation in such a tiny area, just how it's reported by averaging it in.

Are you running the latest BIOS? and AMD chipset drivers? and what are you using to report temperature? I'd suggest getting HWinfo64, if you aren't, and reading the CPU (Tctl/Tdie) and comparing that to the CPU (Average) temp readouts. The difference is the Tctl/Tdie is the hottest spot reported on the CPU at that moment, while the Average is an average of all the sensors.

 

[DadduDavid]

Reported temp spikes like that used to be fairly common with Zen 2 before changes to temp reporting. It's an effect of the tiny 7nm geometry, hot-spot temp sensors and boosting algorithm that lifts individual cores to max clocks from idle for extremely short duration. No amount of cooling, short of sub-ambient, could tame the spikes. At some point after the release of Zen 3, I believe, AMD "fixed" it with updated AGESA and possibly chipset drivers. At least that's what I saw with my motherboard after BIOS and chipset driver updates. I don't suppose that exactly fixed the localized temp spiking during boosts since that's inevitable with high power dissipation in such a tiny area, just how it's reported by averaging it in.

Are you running the latest BIOS? and AMD chipset drivers? and what are you using to report temperature? I'd suggest getting HWinfo64, if you aren't, and reading the CPU (Tctl/Tdie) and comparing that to the CPU (Average) temp readouts. The difference is the Tctl/Tdie is the hottest spot reported on the CPU at that moment, while the Average is an average of all the sensors.

Hello! I am using the latest BIOS version and chipset drivers. I am already using hwinfo64, as you can see from my imgur link. All 3 readings jump to 90c when running cinebench r23 multicore.

 

[drea.drechsler]

... All 3 readings jump to 90c when running cinebench r23 multicore.

Hah...all I saw at first was a glamor shot and BIOS screens, so bailed at that point...the money shot was the last image!

That's normal too. Maybe not so much as Zen 4 CPU's but Zen 3 CPU's are also temp seeking and 90C is the temp they shoot for.

But I do feel 90C is high with a custom loop and running with curve optimizer settings of -30 on all cores. My 5800X3d CO settings aren't that aggressive and it's only getting into the mid-80's in Prime95. I've also set an x5 scalar and +125Mhz boost over-ride. And it is doing this with a 240mm AIO.

The only major settings difference may be your EDC setting. I've found dialing it back from stock (140A) to 120A helps keep a higher clock with better performance. I leave PPT and TDC at stock settings, 142W and 95A respectively. Raising PPT in particular can send temperature higher with no performance improvement in CB20 benchmarks.

How is your system piped? does the water flow to GPU first and then CPU?

 

[CountMike]

Despite that "Custom loop" cooling it could still be a problem with water block having bad contact with IHS or something of that nature, I have just an ordinary AIO 360 and temps stay up to 80c if pushed hard and long, Only CO is set to -30 PBO and everything else is on max producing 4850 MHz all core and 5000MHz single. idle temps up to 35 when in hot room.
There is onemore thing you can try, if BIOS permits, set power limit to 65W, it seems to help forfew degrees les without performance loss.