5800X3D Running 91C on Moderate Load

[R0H1T]

I cant get my X3D to hit 80c even at 140w+

Well that's because you're in Canada.

I won't be able to reseat/repaste the CPU until the 13th, so I was looking for any advice for now. I attached HWINFO screenshots in case I'm missing something. Thanks!

What's your ambient temps? It should definitely run cooler but depending on where you are maybe not by that much!

 

[rusty caterpillar]

My thermal paste is at home and I'm away at college until the 13th I'll try to reseat then.

Your AIO comes with pre applied thermal paste so don't think had any protective plastic film. AM4 brackets of this AIO got only 1 set of holes.
Arctic AIO got 2 sets of holes on the brackets and apparently for Ryzen AM5 you have to use the upper holes and will position the cold plate differently on the CPU for better efficiency. Can't remember if for AM4 is the same but, it could be the case.
I'm happy with my Arctic 280 but, my CPU got a TDP of 65W and can exceed that to max 90W I seen. However the radiator is thicker than other AIOs you have to take that in consideration.
Now Arctic gives 6 years warranty following their issues with the cold plate which apparently they fixed. Time will tell how well Arctic AIO will last..

 

[adilazimdegilx]

I cant get my X3D to hit 80c even at 140w+

View attachment 324655

Hitting 140w with 5800x3d is more impressive than keeping it under 80C I wonder how did you managed that.. +CO? Which program to throttle it to 140w? I can barely get to 110watt @103BCLK with all cores over 4550mhz under load (cb23).

On topic,
There is definitely something wrong with your system. Mine sits at 80C @ said settings above.

 

[rusty caterpillar]

Hitting 140w with 5800x3d is more impressive than keeping it under 80C I wonder how did you managed that.. +CO? Which program to throttle it to 140w? I can barely get to 110watt @103BCLK with all cores over 4550mhz under load (cb23).

Tbh I don't know why hitting more watts on CPU or GPU are considered achievements. Maybe shows stability, but higher wattage not necessarily brings more efficiency of the CPU or GPU or to the whole system, just my cent.

 

[R0H1T]

Hitting 140w with 5800x3d is more impressive than keeping it under 80C I wonder how did you managed that.. +CO? Which program to throttle it to 140w? I can barely get to 110watt @103BCLK with all cores over 4550mhz under load (cb23).

I can get to ~150w on a 2700(non X) so no it's that big of a deal. Probably just to test stability.

Pretty more all Zen chips are artificially (power) limited by AMD & that's generally a good thing for the vast majority of us.

 

[Ferather]

Don't know if this has been mentioned but is the pump connected to the pump header and running at 100%?
I have seen people make that mistake at least 5 times, if no pump header use fan and set 100% always.

 

[Waldorf]

the ryzen offset with some coolers is for AM (4+5), but usually not needed if your not running a "bad" chip (volt wise),
do ocing etc.


@neatfeatguy
right, cause you have tested every cooler on this planet.

enough cooler i installed in the past 20y (incl mine), get screwed down completely (no force, just tight).
and yes, some stuff with larger/longer parts (more common with LC) will flex a bit, but not talking about bending (out of shape).


@Ferather
nope.
most aio regulate pump speed themselves, and some only need temp/pwm info for rpm control, especially those with more than one connector.
ignoring that i was running pumps on a fan header +10y before even a single port was labeled "pump".
most good boards dont even show any physical difference between them.

 

[adilazimdegilx]

Tbh I don't know why hitting more watts on CPU or GPU are considered achievements. Maybe shows stability, but higher wattage not necessarily brings more efficiency of the CPU or GPU or to the whole system, just my cent.

I can get to ~150w on a 2700(non X) so no it's that big of a deal. Probably just to test stability.

Pretty more all Zen chips are artificially (power) limited by AMD & that's generally a good thing for the vast majority of us.

I know, I know. High wattage isn't something to brag about and it was not my point.
It's just in my experience it's not really possible to get those wattages because of the limitations of 5800x3ds. You can set a higher power limit, and stock is 142watts, but it can't really hit those numbers ever. Because you either need higher frequencies or higher voltages to increase power usage (assuming cooling is sufficient). But both those are basically locked in this chip. So even if I set higher power targets and use better cooling cpu wouldn't get past over 105w~ on cb23, ever.
Thats why it's interesting to see a 5800x3d at 140watts.

 

[forman313]

Don't know if this has been mentioned but is the pump connected to the pump header and running at 100%?
I have seen people make that mistake at least 5 times, if no pump header use fan and set 100% always.

On the Strix B550 motherboards Hwinfo usually show Pump RPM if its connected right. In OPs screenshots the CPU-fan is showing a max of ~1100 RPM and that matches the pump speed of the AIO in Quiet mode. So you´re probably right. He needs to connect the pump to the pump-header and verify the pump speed in quiet and normal/extreme.

Most likely there is a contact problem as well. In the screenshots the CPU-temp never drops below 48c .. even at idle.

 

[Princess Garnet]

I'm starting to wonder if there is an issue with the 5800X3D itself. I've been combating 90c temperatures at full load with a Dark Rock Pro 4 and all the fans completely maxed out. It will go from mid 30's to 90 almost instantly as if there was no cooler there at all. I've done everything I can do to mitigate this.
I wonder if some 5800X3D's have IHS issues?

It's not an "issue" if it just spikes there at times, especially on air cooling this can happen (speaking as someone with the same CPU and cooling). It's how Zen 3 and newer tend to behave, especially on air cooling. Although if it's shooting to 90C, staying there, and throttling, then it is an issue. That shouldn't be happening, even on cooling less capable then the Dark Rock Pro 4.

My CPU genuinely runs at 91C the entire time I'm in-game.

Definitely not normal! I'm typically in the upper 50s to 60s in-game, but it goes into the 70s too when more cores go under load if I'm generating/loading a lot of chunks at a higher render distance.

Spikes to near 80C to 90C, or not-exactly-the-same-as-mine temperatures, would be one thing. Staying at 90C though? When the game is largely lowly-threaded? You have a cooling issue that needs investigated. You need to reseat/repaste it and see where that gets you.

 

[Waldorf]

@forman313
the corsairs set speeds based on the profile selected in the icue sw.

 

[freeagent]

Well that's because you're in Canada.

Nah.. might be chilly outside, but we have a box in the basement that makes fire, and has a fan to blow the heat from the fire around

I can get to ~150w on a 2700(non X) so no it's that big of a deal. Probably just to test stability.

I can do 260w with my 5900X, but X3D turns your computer into a Dell.

I know, I know. High wattage isn't something to brag about and it was not my point.
It's just in my experience it's not really possible to get those wattages because of the limitations of 5800x3ds.

Linpack Xtreme, the 10GB load.

 

[agent_x007]

Did you pre-spread thermal paste before mounting the cooler ?

 

[parufy]

Well that's because you're in Canada.

What's your ambient temps? It should definitely run cooler but depending on where you are maybe not by that much!

my homepod mini sits next to my pc and it usually says it’s ~80°F indoors (that thing ain’t lyin). there’s 4 of us in a kinda small dorm but all of these tests were taken when I was the only one home.

Don't know if this has been mentioned but is the pump connected to the pump header and running at 100%?
I have seen people make that mistake at least 5 times, if no pump header use fan and set 100% always.

the pump is at 100% duty in the bios and I am actually not allowed to change it.

i just remembered that i had the exact same problem and the solution for it back when both the cooler and zen 3 were brand new.
the CoolIt Coldplates from the XT (had two of them) is so convex that it makes really bad contact on ryzen.
both of my H150i Pro XT were really hot on my 5900X because the coldplate is specifically designed for the concave intel IHS.
even found my photo of my finished system. and i remember how unhappy i was with this AIO.

View attachment 324689
View attachment 324690

Hoping I don’t need to get a new cooler but thanks for the insight

On the Strix B550 motherboards Hwinfo usually show Pump RPM if its connected right. In OPs screenshots the CPU-fan is showing a max of ~1100 RPM and that matches the pump speed of the AIO in Quiet mode. So you´re probably right. He needs to connect the pump to the pump-header and verify the pump speed in quiet and normal/extreme.

Most likely there is a contact problem as well. In the screenshots the CPU-temp never drops below 48c .. even at idle.

What about the third screenshot, second to last dropdown? Says ~2500RPM. My fans max out ~1200-1400RPM.

@forman313
the corsairs set speeds based on the profile selected in the icue sw.

I’ve been seeing an iCUE unexpected crash error message like every time I unlock my PC and haven’t cared to check ;( I’ll check.

Uninstalled iCUE>rebooted>reinstalled iCUE>still crashing on startup>run sfc /scannow>rebooted>crashes stopped?

Looking in iCUE, it reported the pump running at ~2500RPM like HWINFO stated. I ran Minecraft again and my CPU runs around 86C. The room temp is about 10F cooler than yesterday, so that makes sense. It's like 50-60F outside, but even with the windows open it gets warm in here.

 

[Ferather]

~4000 RPM pump here with also a 360 AIO, I would check the specs online to make sure the pump is meant to be ~2500 RPM, also check in the bios for DC (3 pin) not PWM (4 pin).

You can focus on other possible issues if the ~2500 RPM is normal for that pump, and the entire cooler unit is rated for the CPU.

 

[parufy]

~4000 RPM pump here with also a 360 AIO, I would check the specs online to make sure the pump is meant to be ~2500 RPM, also check in the bios for DC (3 pin) not PWM (4 pin).

You can focus on other possible issues if the ~2500 RPM is normal for that pump, and the entire cooler unit is rated for the CPU.

~2500 is normal. I set it to max speed and it jumped to ~2800RPM which didn't do much for end results.

 

[Waldorf]

@Ferather
how many pins a fan port has, has NOTHING to do with how its controlled.

 

[Ferather]

DC uses varying voltage, PWM full 12v and speed signal. Still its in my bios as an option to choose between on AIO_Pump header, may as well be the right option.

 

[Redwoodz]

3D cache is probably getting hammered during those Minecraft sessions. Test with something like Cinebench and compare. Remember you are dealing with sensors and software that can read wrong also. Your ambient temps are up there so it's not out of the question you could have a sensor slightly off with those perameters you've shown. An 85c temp would be normal under those same conditions.

 

[Waldorf]

@Ferather
as long as its on auto, the corsair dont care much about it.
asus 550:
1 x 4-Pin AIO_Pump header
1 x 4-Pin CPU fan header
1 x 4-Pin CPU_OPT fan header
3 x 4-Pin Chassis fan headers

short of the lowest tier, virtually all x5xx boards i used for builds had all 4 pin ports, doing DC & PWM,
and all +200$ boards even coming with same output wattage on all, so it wont matter if its a fan/pump or anything else.


and no, pwm has nothing to do if speed signal or not, just that it pulses on/off, versus lowering V (on DC),
as i have both types of fan working without speed signal feedback (cable extension).

 

[AMF]

Hey yall,

I spontaneously checked my temps on HWINFO while playing modded Minecraft and my CPU temps were ludicrously high. It didn't indicate any throttling, but it seemed excessive for the workload. So, I did some investigation into lowering my temps. My motherboard was running a BIOS from April 2022 (when I upgraded to the 5800X3D), so I updated it to the latest version, enabling PBO2. I reapplied my fan curves (100% fan duty at 90C) and rebooted. Same temps; pinned at 91C. Seeing this, I set the curve optimizer to -30, PPT to 100, TDC to 70, and EDC to 100. Rebooted, and... same temps. The clocks increased a solid 200MHz, though.

I won't be able to reseat/repaste the CPU until the 13th, so I was looking for any advice for now. I attached HWINFO screenshots in case I'm missing something. Thanks!

Relevant specs:
CPU: Ryzen 7 5800X3D
Cooler: Corsair H150i RGB PRO XT (top-mounted exhaust)
Motherboard: ASUS B550-F GAMING
Case: Lian Li PC-O11 Dynamic (all 9 fans)
GPU: XFX MERC 310 RX 7900 XTX
PSU: Corsair RM850x SHIFT

disable l1 l2 hardware prefetcch

 

[parufy]

disable l1 l2 hardware prefetcch

Could you elaborate on what that does?

 

[AMF]

Could you elaborate on what that does?

its a prefetch for the cache on your cpu . u may take a slight latency penalty disabling it but it will stop some heat generation . u will gain some speed on cpu though. disable it in your bios and try it if u dot like it u can turn it back on

 

[chrcoluk]

Seen some recent information on https://www.techpowerup.com/download/amd-ryzen-clock-tuner-ctr-1usmus/

People with bronze rated CPU's having massively different characteristics on their chips vs gold/platinum rated samples, looks like AMD has the same wild variation as Intel.

The OP does seem to have a cooling issue, but I also wonder if his chip is a bronze rated sample.

 

[parufy]

Hey again, everyone!

I spent yesterday putting my desktop setup back together, but I didn't want to spend all day working on my PC. So, today I got working on the issue itself. Silly story, I had an extra AIO (EVGA CLC 360) from when I watercooled my old GPU. Having seen @GerKNG's message, I decided to go all in and swap out my cooler for the EVGA one. When I removed the Corsair cooler, I may have smeared the thermal paste a bit. Regardless, the paste showed patches of clear liquid. I'm fairly sure that means the electrically conductive parts of the paste were no longer present and only the binding liquid remained.

I pre-spread the thermal paste on my CPU before mounting the EVGA cooler. It uses four screws instead of two and I tightened it only until the screws started resisting. I also ensured the pump was on max speed in the BIOS. For reference, my Homepod says my room is at 65F with the windows open. I ran a 10 min Cinebench benchmark then played some Minecraft, and my CPU never went past 77C. Consequently, my problem is solved. I know this solution might not work for everyone else, but those with a Corsair cooler should attempt to reseat the cooler and DO NOT tighten the screws down all the way. Thank you all for your help.