Strange CPU overheat issue.

[t33to]

So I've got the DeepCool Castle 240RGB on a Ryzen 3600 inside of an h500p-mesh case.

 

Running Prime95 the CPU hovers around 60c. This is all threads and cores at 100%, each core averaging 4150mhz.

 

It will run like this, nice and steady as you would expect, when about 10 minutes into the test, all of a sudden it will just shoot up to 80c+ and then shortly after that it auto shuts down.

 

What could be causing this? The only thing I can think of is that I'm using shitty thermal paste. The AIO came with paste itself but I'm guessing this cooler is really old as the paste basically came out like rubber out of the tube, it wasn't pastey at all and couldn't be spread. So I went to London Drugs and got some of their thermal paste and it's pretty runny. Almost watery. I've got some Noctua thermal paste on the way. I still don't understand how the thermal paste could be the issue though.

 

Pump is steady at 2500rpm, both 120mm fans on the rad are steady at 1800rpm. Both fans are running off of one fan header, connected to a splitter.

[Nord]

Apart of the thermal paste it could also be an uneven mount.
 

Also depending on what you are using to measure the temps, it may “averages” them badly. The average core temperature of 12 (theoretical) cores can easily be off by a huge amount if it's just one single core that's heating up.
To be sure check single core temperatures and also make sure you are reading the core temps and not the TCase / CPUTin temp.

I’m not into water cooling so take what I say next with a grain of salt:
Perhaps the loop is bad, pump is broken or too weak or the water is contaminated.
If I recall correctly, as it takes some time for water to heat up, from a cold start you could easily go 10+min in prime95 even w/o a pump until everything heats up and the core temps rise - or rather shoot up.

[t33to]

12 hours ago, Nord said:

Apart of the thermal paste it could also be an uneven mount.
 

Also depending on what you are using to measure the temps, it may “averages” them badly. The average core temperature of 12 (theoretical) cores can easily be off by a huge amount if it's just one single core that's heating up.
To be sure check single core temperatures and also make sure you are reading the core temps and not the TCase / CPUTin temp.

I’m not into water cooling so take what I say next with a grain of salt:
Perhaps the loop is bad, pump is broken or too weak or the water is contaminated.
If I recall correctly, as it takes some time for water to heat up, from a cold start you could easily go 10+min in prime95 even w/o a pump until everything heats up and the core temps rise - or rather shoot up.

Thanks. I'll take all of that into account.