Crazy-high Ryzen 3900X temperatures (95c immediately)

hititnquitit said:
Check the hoses. If one is hot and the other is cool you likely have a dead pump or theres gunk in the loop clogging up the block or one of the hoses.
Click to expand...

Thanks, this is the case with my 115i pro. Luckily it is within warranty so corsair is fine with replacing it but I'm not 100% on how expensive that will be considering I'm in Canada. I was hoping this wasnt the case as I just swapped out the old 1700 and assumed it was the processor causing the problems ( gigabyte gaming 5 x 370 board)

Centauri how were your temperatures after?
 
You would think with all the stupid rgb lights they put on coolers (especially aio's) that they would include a simple pump impellor rpm sensor. I'm not aware of any that do....much less any that feed that info back to the motherboard via the rpm signal letting the motherboard detect the failure and shut down automatically itself without needing the system to go into overheat protection first.

I've had 1 aio out of 5 suffer that same fate. Luckily, cpu's these days are pretty good about saving themselves from perm damage via overheating.

edit: though i'm not sure if the motor failing still incorrectly registers an rpm ....or if the bios doesn't have a failsafe you can set to emergency poweroff when fan rpm == 0 ... My idea above was more inline with if the impellor for some reason fails to spin but the motor is still going.
 
Darth Ender
upload_2019-8-28_17-13-53.png


like this?
 
Last edited:
I know they report pump rpm via the rpm of the brushless motor that usually powers them.

But it's not directly measuring the rotation of the impeller blades. And I dont happen to know what that rpm signal looks like when the pump fails in some way (be it total power failure or impellor rotation failure).

I dont recall there being any options in the bios either on treating 0rpm on a given fan header (whether it's the aio pump header or any cpu fan header) as an emergency poweroff. I'm sure third party software exists in the OS to do such a thing, but it would be nice to know that the bios can be setup to do it so there's no dependency on the OS.
 
Darth Ender said:
I know they report pump rpm via the rpm of the brushless motor that usually powers them.

But it's not directly measuring the rotation of the impeller blades. And I dont happen to know what that rpm signal looks like when the pump fails in some way (be it total power failure or impellor rotation failure).

I dont recall there being any options in the bios either on treating 0rpm on a given fan header (whether it's the aio pump header or any cpu fan header) as an emergency poweroff. I'm sure third party software exists in the OS to do such a thing, but it would be nice to know that the bios can be setup to do it so there's no dependency on the OS.
Click to expand...

Pretty sure this is plugged into the cpu fan header, I'll have to check =) (there are options for alarms, don't know about shut downs on cpu fan header)
 
I'm usually hovering around 35C-40C with an ambient room temp of 23C. I do have a hefty amount of air movement though. Corsair H150i Pro, 6 Corsair ML120 RGB Pro's for intake and 3 Corsair ML120 RGB Pro's as push exhaust out the top of the case. The chip is idling at 4GHz.
 
I run 40-50 with the CPU reading 4.25Ghz and 69 under 100% load. H100i cooler running balanced pump speed normal or extreme when gaming.
 
Jamie Marsala said:
I run 40-50 with the CPU reading 4.25Ghz and 69 under 100% load. H100i cooler running balanced pump speed normal or extreme when gaming.
Click to expand...

That's pretty good for a 100% load. Usually, if I run Aida64 stess test, I get way higher than that (80s) and my case is a wind tunnel. I'm running the H100i Platinum SE in push-pull with a fan curve tracking the CPU temperature that tops out at 100% fan speed at 90 degrees.
 
all fully loaded temps are not created equal. The full load the most stressful prime95 stress test can provide (they have multiple modes with a balanced as default, as opposed to a max stress) will provide a massively different level of power use in the cpu than most other benchmarks that center around some actual real functional work. Both will show up as using 100% of all cores, but the temps can be 20+C different.

Pretty much anything past 4.1Ghz on the 3900x will lead to ever diverging relationships between power output and performance increase. If you're willing to sacrifice literally 1 or 2 % of peak potential performance, you can keep your temps a very real 10-20C cooler than the guy who pushes for 4.3-4.6Ghz

So it's very easy to throw around whatever numbers you want and not really be lying. But that's nothing new when it comes to comparing temps.
 
Darth Ender said:
You would think with all the stupid rgb lights they put on coolers (especially aio's) that they would include a simple pump impellor rpm sensor. I'm not aware of any that do....much less any that feed that info back to the motherboard via the rpm signal letting the motherboard detect the failure and shut down automatically itself without needing the system to go into overheat protection first.

I've had 1 aio out of 5 suffer that same fate. Luckily, cpu's these days are pretty good about saving themselves from perm damage via overheating.

edit: though i'm not sure if the motor failing still incorrectly registers an rpm ....or if the bios doesn't have a failsafe you can set to emergency poweroff when fan rpm == 0 ... My idea above was more inline with if the impellor for some reason fails to spin but the motor is still going.
Click to expand...

My corsair h110 has pump speed sensor.

Your pump speed sensor port on your motherboard will in fact trigger an alert or run a shutdown command if it fails to meet a threshold.

Gigabyte software does this.
 
pump speed sensors that come from the motor that drives the pump almost definitely are just halls effect type sensors that monitor the motor. It should do a good job at detecting impellor speed but if the shaft is spinning but the blades aren't due to corrosion causing the blades to no longer be tightly coupled to the shaft...for instance, then it'll look like the pump is working at a given rpm but the blades aren't actually pushing fluid. You'd need a separate sensor watching the blades ..or a flow meter that is monitored and outputs a pwm signal like a fan does that the motherboard can monitor.

In any case, this person's pump failed and the motherboard didn't trigger a failsafe condition. So it's apparent that such features are either opt-in or not always available on boards.
 
kamikazi said:
That's pretty good for a 100% load. Usually, if I run Aida64 stess test, I get way higher than that (80s) and my case is a wind tunnel. I'm running the H100i Platinum SE in push-pull with a fan curve tracking the CPU temperature that tops out at 100% fan speed at 90 degrees.
Click to expand...

I just bought the test suite last night. Running the tests individually I was now hitting 90. Run them all at once and I must have hit the thermal limit since the computer shut down and left just my fans and pump on. I am thinking the H100i Pro is either not enough cooling or I did not put enough paste on. I will pull it today and see what is up and if there is enough paste then I will switch to the Prism cooler and see how that does. I had my pump on Extreme speed and the fans should have been at 100% at anything over 75 degrees. The software says the cooler temp is 30 degrees at idle when the processor is reading 40 so I assume something is up. My 2600 read pretty close temp wise.
 
So I finally added the Wraith Prism - idles at about 39 and max i've seen the temps go to is a multicore cinebench run at 80.

Huge difference. At least corsair is replacing the old cooler for "free" though.
 
You must log in or register to reply here.
Back
Top